This post summarizes a Founders Pledge shallow investigation on direct communications links (DCLs or "hotlines") between states as global catastrophic risks interventions. As a shallow investigation, it is a rough attempt at understanding an issue, and is in some respects a work in progress.
Crisis-communication links or “hotlines” between states are a subset of crisis management tools intended to help leaders defuse the worst possible crises and to limit or terminate war (especially nuclear war) when it does break out. Despite a clear theory of change, however, there is high uncertainty about their effectiveness and little empirical evidence. The most important dyadic adversarial relationships (e.g., U.S.-China, U.S.-Russia, Pakistan-India, India-China) already have existing hotlines between them, and forming new hotlines is an unlikely candidate for effective philanthropy. Along with high uncertainty about hotline effectiveness in crisis management, the highest stakes application of hotlines (i.e., WMD conflict limitation and termination) remains untested, and dedicated crisis-communications channels may have an important fail-safe role in the event of conflict.
War limitation- and termination-enabling hotlines have high expected value even with very low probability of success, because of the distribution of fatalities in WMD-related conflicts. Importantly, it appears that existing hotlines — cobbled together from legacy Cold-War systems and modern technology — are not resilient to the very conflicts they are supposed to control, and may fail in the event of nuclear war, electro-magnetic pulse, cyber operations and some natural catastrophic risks, like solar flares. Additionally, there are political and institutional obstacles to hotline use, including China’s repeated failure to answer in crisis situations.
Philanthropists interested in crisis management tools like hotlines could pursue a number of interventions, including:
- Funding work and dialogues to establish new hotlines;
- Funding work and dialogues on hotline resilience (including technical work on hotlines in communications-denied environments);
- Funding more rigorous studies of hotline effectiveness;
- Funding track II dialogues between the U.S. and China (and potentially other powerful states) focused on hotlines to understand different conceptions of crisis communication.
We believe that the marginal value of establishing new hotlines is likely to be low. The other interventions likely need to be sequenced — before investing in hotline resilience, we ought to better understand whether hotlines work, and what political and institutional issues affect their function. Crucially for avoiding great power conflict, we recommend investing in understanding why China does not “pick up” crisis communications channels in times of crisis.
Acknowledgments: I would like to thank Tom Barnes, Linton Brooks, Matt Lerner, Peter Rautenbach, David Santoro, Shaan Shaikh, and Sarah Weiler for helpful input on this project.
Thomas Schelling first suggested the idea of a direct communications link between the United States and the Soviet Union in 1958, and the idea was popularized in outlets like Parade magazine. Although early attempts were made at implementing such a link (e.g. in early 1962), the need for such a dedicated communications channel between the United States and Soviet Union became pressingly clear during the Cuban Missile crisis, when Kennedy and Krushchev communicated through “clumsy” and slow traditional communications channels. Officials at the Soviet embassy in Washington later recalled that even their own communications with Moscow used slow and unreliable means; when communicating by telegraph, “we at the embassy could only pray that he [the messenger] would take it to the Western Union office without delay and not stop to chat on the way with some girl.” At one point, Kruschev even replied to Kennedy via a broadcast on Radio Moscow, because diplomatic channels were too slow. The leaders decided on a hotline agreement, which went into force on August 30, 1963, establishing a teletype (not telephone) link between Moscow and Washington.
Early versions of the U.S.-Moscow hotline were fragile and insecure — in one case, a farmer in Finland accidentally plowed through the cable in 1965 — and several improvements and updates have been made since then, including switching to satellite communications in 1971, adding fax in 1984, and email in 2008.
Since then, a number of other “hotlines” have been established, which we have mapped in Google Earth Pro. Some of these states have multiple hotlines, and they are implemented at different levels (head-of-state, military-to-military, etc). Moreover, some of them may be inactive. Previously established hotlines include:
- U.S.-Russia hotline (1963)
- France-Russia hotline (1966)
- UK-Russia hotline (1967)
- North Korea-South Korea hotline (1971)
- U.S.-China hotline(s) (1998)
- China-Russia hotline (1998)
- India-Pakistan hotline (2004)
- South Korea-China hotline (2008)
- India-China hotline (2010)
- Vietnam-China hotline (2012)
- Taiwan-China hotline (2015)
- U.S.-India hotline (2015)
- Greece-Turkey hotline (2020)
- Japan-China hotline (2022, defense-level)
Notably, we were unable to find any attempt to fully map the existing bilateral hotlines or their degrees of removal. We have illustrated our understanding of the dyadic hotline relationships in the following table (which is simplistic in not distinguishing between allies and adversaries, but provides a rough overview):
Allies have other means of communication, and may have a lower need for a hotline. All else equal, however, it becomes clear that although the largest militaries and great powers have established hotlines, many of the world’s nuclear-armed states remain unconnected via crisis communications channels. This could be problematic for the resolution of regional conflicts that have the potential to spark great-power war.
To prioritize for the establishment of multiple hotlines, philanthropists and policymakers can use a simple prioritization framework:
- Prioritize larger size of damage conditional on conflict (e.g., a U.S.-Russia war may be far more disastrous than a U.S.-DPRK war);
- Prioritize higher probability of conflict (i.e., hotlines between allies may be less pressing than between adversaries, all else equal).
This could be easily quantified (e.g. by forecasting conflict probability and using military spending or nuclear arsenals as a proxy measure for military power) into a multiplier framework. As the following sections explain, however, there is high uncertainty around the effectiveness of the crisis-management function of hotlines, especially in light of some states’ apparent divergent understandings of the purpose of hotlines, and the most important states already have hotlines.
Hotline Theory of Change
Crisis-communication hotlines between states have a three-part theory of change:
- Crisis Management — Hotlines can provide a rapid and direct communications link between leaders to facilitate information exchange, sharing of intentions, and cooperation to de-escalate crises.
- War Limitation/Escalation Control — In the event of war, hotlines can provide a channel to communicate the intended limits to that war and thereby control escalation.
- War Termination — Hotlines provide the means to communicate a desire or intent to terminate a conflict (e.g., in the midst of a nuclear war, both sides can outline the conditions for peace). We are following convention in listing termination as a separate problem from limitation, but believe they face similar practical challenges, and hence treat them in only one section below.
Additionally, there are indirect benefits to a hotline, including as a confidence-building measure to seek to increase trust and mutual understanding. As the Cold War strategist Thomas Schelling put it, “the mere establishment of the hotline was itself to some extent a confidence-building step, although the main purpose was to facilitate restoration of legitimate confidence in a crisis,” because this establishment communicated that both sides took the risk of accident and unintended escalation seriously.
There are game-theoretical reasons for believing that hotlines could be important tools for de-escalation. Tacit bargaining — that is, “bargaining in which communication is incomplete or impossible,” in Schelling’s words — is extremely challenging and can lead to suboptimal outcomes. Two (or more) nuclear powers in conflict can be modeled as a complicated iterated game, where the choices are “escalate” and “de-escalate,” and mutual de-escalation is usually in the ultimate interest of all players. These powers might converge on a focal point where both decide to de-escalate at a certain point, but such a focal point might be difficult to find tacitly when there are no clear qualitative differences, or the focal point might be at a place that proves catastrophically destructive to all of humanity.
Allowing for explicit bargaining would enable the powers to find arbitrary focal points that avoid the worst possible outcomes. In Schelling’s words: “There is no assurance that the next war, if it comes, will find mutually observed limits in time and of a sort to afford protection, unless explicit negotiation can take place.” Hotlines can help to facilitate this explicit negotiation. The following sections discuss the three hotline applications. We find that there is high uncertainty over whether hotlines have counterfactually led to de-escalation in crises, argue that crisis management hotlines may be more redundant now than they were during the Cold War, and point to China’s failure to use hotlines in crisis situations with the United States as a major obstacle to their effectiveness in a future conflict.
Application 1: Crisis Management
High Uncertainty on Effectiveness
As discussed above, crisis-communication hotlines between states have a strong theory of change. Moreover, some leading decision-makers have praised hotlines for their usefulness in crisis. For example, former Secretary of State Caspar Weinberger said that the hotline “has proved invaluable in major crises.”
Unfortunately, data on hotline use is sparse, and there is high uncertainty on their effectiveness. As noted by the Arms Control Association, “an official listing of the instances when the states used the hotline never has been released to the public.” Selective declassification and reporting bias where primary sources are unavailable mean that any database of hotline uses may be systematically biased in one direction or another. We have drawn base rates from the Six Day War, as all hotline messages exchanged during the crisis have been declassified (see Appendix: Six Day War Hotline Use for greater detail). In this conflict, we code exchanged messages as:
- 85.7% Crisis Management;
- 4.8% Threats/Escalation;
- 9.5% Mixed/Unclear.
Note that using the hotline to convey threats (in ~5% of cases for this crisis) is not necessarily bad; it still communicates information about intent and interests to one’s adversary, which can be useful in understanding whether apparently benign actions could be perceived as escalatory. This uncertainty also runs the other way, however, as it is not obvious that intended crisis management leads to actual crisis management.Most fundamentally, the lack of a counterfactual complicates any analysis of even declassified hotline uses.
At times, the hotline was apparently ignored when it was needed most. For example, during the infamous 1983 Able Archer incident, where the USSR misinterpreted NATO wargames as an apparent preparation for attack, there was no known use of the Washington-Moscow link to communicate about the misunderstanding. One may argue that the proliferation of hotlines around the world is itself evidence of its effectiveness, and that perhaps the most effective uses remain classified. However, as one assessment of hotline diplomacy explains, an alternative explanation may be that hotlines are seen as “prestigious” tools of diplomacy that establish an actor’s importance, rather than as useful crisis communications links. Alternatively, hotlines may spread because of the mere perception of effectiveness; states may view hotlines as a relatively simple way to reduce risk, even if there is little or no evidence of the effect.
U.S.-China Crisis Communication Weaknesses
China has the most hotlines of any state, but is an unreliable hotline partner. For example, after the establishment of an initial presidential-level communications link in 1997, Chinese leaders did not respond to repeated U.S. contact attempts during the 2001 Hainan Island incident. In this incident, Chinese fighter jets got too close to a U.S. spy plane conducting routine operations, and the U.S. plane had to make an emergency landing on Hainan Island. The U.S. plane contained highly classified technology, and the crew destroyed as much of it as they could (allegedly in part by pouring coffee on the equipment) before being captured and interrogated. Throughout the incident, the U.S. attempted to reach Chinese leadership via the hotline, but were unsuccessful, leading U.S. Deputy Secretary of State Richard Armitage to remark that “it seems to be the case that when very, very difficult issues arise, it is sometimes hard to get the Chinese to answer the phone.” This kind of incident is precisely the moment at which a crisis-management hotline would in theory be useful, and the lack of response may be a reason to update down on the probability that a hotline would help to manage future U.S.-China crises.
More recently, Kurt Campbell, the Biden administration’s Coordinator for Indo-Pacific Affairs in the National Security Council has lamented that “In the past, the hotlines that have been set up have just rung, kind of endlessly in empty rooms,” in China. Nonetheless, U.S. defense officials are actively pursuing stronger direct communications links between the two countries.
In short, there is high uncertainty about hotline effectiveness, and some evidence that China routinely ignores hotlines in crises. The lack of counterfactual evidence makes evaluating hotline success especially difficult, and there are potential confounders — for example, improved relations between states could lead to both hotline establishment and the peaceful resolution of crises, without a causal relationship between hotlines and crisis resolution. Multi-player experimental wargames may help provide some clarity, but as far as we are aware, there are no such games testing the effect of hotline presence or absence (or uncertainty about hotline resilience).
Without strong empirical evidence, and without being able to simulate counterfactuals, it is difficult to know whether the absence of another Cuban Missile Crisis — or even the absence of a nuclear war — since the establishment of the hotline is at least partly causally attributable to the existence of hotlines. It could be, however, that most hotline value is concentrated in a small handful of cases. One possible approach is to look at moments in history that were “contingent” in MacAskill’s sense of the word: “Contingency represents the extent to which a state of affairs depends on a small number of specific actions.” Some nuclear “close calls” appear to be highly contingent moments, where the superpowers plausibly could have teetered over the edge of war but for a few specific actions. In at least one nuclear close call, the hotline appears to have been instrumental in avoiding a crisis, when on 28 December 1984, the Soviet Union used the Washington-Moscow hotline to let the United States know about a mistaken missile launch which appeared to be on a course to Hamburg, but crashed in Finland. There is little historical evidence on this case, however, making it difficult to evaluate the claims. More recently, during the March 2022 mistaken missile launch from India to Pakistan, which led to Pakistan preparing for a retaliatory strike, the hotline between the two countries was reportedly not activated.
Given the dearth of evidence, and the lack of counterfactuals, there is high uncertainty on the effects of hotlines. Notably, high uncertainty is not per se a reason for inaction, especially if the upside is sufficiently large. As the following section explains, however, hotlines appear to be more redundant in 21st century peacetime scenarios than they were during the Cold War, and there are already established hotlines for the most important dyadic relationships.
Advances in communications technology since the 1960s have diminished the hotline’s unique utility as a communications channel. Since 1963, the proliferation of cell phones, and the ease of long-distance communication may have addressed some of the concerns that initially led to the establishment of the hotline (the slowness of diplomatic channels). One example of leaders de-prioritizing hotline communications and opting for cell-phone communications can be seen by then-President Donald Trump’s exchange of cell phone numbers with North Korean leader Kim Jong Un. The explanation Trump gave mirrors the justification often given for hotlines: “I can now call him. I can now say, ‘Well, we have a problem.’ I gave him a very direct number. He can now call me if he has any difficulties, I can call him.” In fact, one can imagine an unofficial network of cell-phone “hotlines” — leaders and staffers meet at diplomatic dialogues, they exchange phone numbers, they can communicate with each other in crises, targeting the communications to the appropriate level of seniority.
Nonetheless, hotlines remain in use, and the U.S.-Soviet hotline has been used “when faster channels of diplomatic communications have been at the disposal of both superpowers.” Scholars have theorized that the reasons for this are that the hotline conveys urgency, importance, intent to cooperate, and other “symbolic” attributes. Other explanations have focused on the hotline as a facilitator of “alter-casting” — taking on the role of a trusting adversary in a crisis — based on “symbolic interactionist role theory,” although the main advocates of this explanation of hotline effectiveness (Eszter Simon and Agnes Simon) provide little evidence for the underlying theory, and appear to select their case studies based on the dependent variable (cooperation in crisis).
There may be additional advantages to the hotline (e.g. in signaling the seriousness of a crisis), but its unique value in crisis management is no longer its speed or directness. Security and spoofing are an ongoing concern, but widespread adoption of end-to-end encryption, as in dedicated apps (Signal), may make it easier to trust the message.
Most importantly, as Table 1 illustrates, the largest nuclear powers, and those most likely to go to war, already have hotlines in place. There are hotlines between the three great powers, the U.S., China, and Russia; there is a hotline between one of the most important regional nuclear rivalries, India and Pakistan; and between India and China, two rising powers with an ongoing border dispute. We have illustrated these connections in the following graph, where the area of each country’s square is proportional to the size of its nuclear arsenal (or deployed arsenal, where known):
In short, the most important hotline links have already been established. Moreover, even if we believe new hotline links are important, the sequencing of interventions suggests that philanthropists should target other issues — including non-use, non-response, and resilience — first. This includes the apparent misunderstandings between the United States and China over the use and non-use of hotlines. As we explain in the following sections, for example, existing hotline links are brittle in wartime scenarios. Rather than establish new links between minor powers that are unlikely to go to war (e.g., the UK and Israel), therefore, effective philanthropists should focus on existing problems with hotlines.
Applications 2 and 3: War Limitation and Termination
Hotlines can act as a war-limitation tool, in addition to their crisis-management function. Senior U.S. leaders have in the past identified the hotline as a crucial component of this, as when Secretary Schlesinger explained the benefits of maintaining “continued communications with the Soviet leaders” during a war, and describing “precisely and meticulously the limited nature of our actions.” Notably, this does not necessarily require belief in the precise controllability of escalation, but merely that it is possible at some point to want to communicate “We are not going full-scale; please don’t respond full-scale.” As Ash Carter explained in 1987, “some thought needs to be given to communications between the superpowers, since terminating a nuclear war before it escalates to all-out exchanges is a goal of U.S. strategy.” Moreover, hotlines have been identified as a crucial method — in fact, the only officially-acknowledged method — for terminating a nuclear war.
This highest stakes application of hotlines (i.e., WMD conflict limitation and termination) remains untested, and dedicated crisis-communications channels may have an important fail-safe role in the event of conflict.
To understand why the existence of resilient hotlines may be useful, it is important to understand two points:
- Resilience — Normal communications like cell phones, radio, and the Internet (but also potentially including brittle hotlines — see below) may be likely to break down in the event of war. If leaders want the option of limiting war, they may wish to build more resilient hotlines.
- Distribution of Hotline Value — Because of the large range of possible outcomes in the event of war, and because communicating escalation is easier than communicating de-escalation, the majority of hotline value may lie in war limitation and termination, as discussed below.
Importantly, existing hotlines — cobbled together from legacy Cold-War systems and modern technology — may not be resilient to the very conflicts they are supposed to control or terminate, and may fail in the event of nuclear war, electro-magnetic pulse (EMP), cyber operations and some natural catastrophic risks, and the intentional severing of communications links. These are the same risks that all nuclear command, control, and communications systems (NC3) face: as a “system of systems,” that combines modern and older technology, they are brittle and vulnerable. The U.S. government has been actively soliciting input for NC3 modernization since 2018. Simply put, NC3 systems are a “patchwork” and “The DCL [Direct Communications Link] is not designed to survive or function in a war environment.” (It should be noted, however, that details about the technical specifications of hotlines and about the resilience of related communications networks are not public, and Ball’s 1991 assessment may be misguided or outdated.)
While it may appear strategically irrational to directly attack a hotline, the infrastructure may be an accidental casualty of a nuclear war, especially a large-scale war where nuclear powers’ capitals are under attack. This has led the Australian defense scholar Desmond Ball to point out the “irony” of current hotline systems: in situations where the hotline is needed most — to prevent an ongoing nuclear war from becoming a cataclysmic thermonuclear exchange — the hotline may be least reliable, even though “it is from this point on that casualties are likely to be several times greater than for the phase when the DCL [direct communications link] was available.”
The full list of vulnerabilities is beyond the scope of this investigation. Potential vulnerabilities of existing hotlines and related NC3 capabilities include:
- Vulnerability of systems or their components to cyber operations;
- Targeting prioritization of communications system in the event of war;
- Supply chain insecurity on the level of chips and software in complex systems of systems;
- Vulnerability of involved space-based systems to ASAT attacks;
- Proximity of hotline terminals to high-value targets (i.e., Moscow and Washington), and probability of collateral damage;
- Vulnerability to electromagnetic pulse attacks;
- Vulnerability to solar storms;
- Vulnerability to large magnitude explosive volcanic eruptions;
- Potential for incapacitation of human network operators in the event of war.
In some cases, the very event that incapacitates hotlines would make hotlines especially useful. This is obviously true for nuclear war, but also for some natural events. In 1967, for example, solar flares knocked out U.S. Air Force early warning radar stations, leading to the interpretation that the Soviet Union had incapacitated these radars in preparation for a first strike. In the words of the Future of Life Institute, “Nuclear bombers were prepared to take flight, but just in time, the recently established Solar Forecasting Center was able to get a bulletin into the hands of a commanding officer showing that a solar flare - and not the Soviets - had knocked out the radar systems.”
The Distribution of Hotline Value
Ball’s above point about the casualties “from this point on” being “several times greater” is crucial to understanding the expected value of a hotline for war limitation and termination; from the perspective of minimizing deaths, an all-out thermonuclear war is orders of magnitude worse than a limited nuclear war. In other words, the distribution of cost in nuclear war makes escalation control a high-leverage point of intervention.
A Fermi estimate helps to illustrate the value of escalation control. (For illustrative purposes, we calculate this in terms of current-generation value, rather than longtermist and existential risk-reduction value, which may depend on the severity and length of nuclear war-induced climate effects.) Consider two scenarios:
- First, a limited nuclear war that kills 1 million people, and
- Second, an all-out thermonuclear war that kills 1 billion people via a mixture of direct effects and nuclear winter.
- For the purposes of the estimate, suppose that the probability that any given nuclear war escalates to a cataclysmic nuclear-winter-causing war is 1-in-5, and that 4-in-5 wars remain “limited” – though still catastrophic — on the order of 1 million dead, i.e. .
- Next, consider that a non-resilient nuclear hotline has zero escalation-control value (because it will not work in the event of war), such that .
- A resilient hotline, on the other hand, might reduce this risk, even if only by a little bit — for example, in one out of one thousand scenarios, the nuclear hotline prevents escalation — such that .
- We further assume that there is a 1% annualized chance of nuclear war.
Given these Fermi-estimate assumptions, we can calculate that the annual expected lives saved through this hypothetical small increased resilience — again, a mere 1-in-1,000 decrease in escalation risk from 20% to 19.98% — would be 2,000 lives per year.
It is, however, possible that uncertainty about communications resilience is itself valuable in deterring conflict. Such uncertainty may itself make a party in a conflict reluctant to escalate beyond a certain point, because they would have greater difficulty communicating their intentions and threats. As discussed below, whether such a deterrent effect exists, and whether it outweighs the benefits of being able to communicate explicitly, could be tested partly in an experimental wargame. In short, hotline resilience is a crucial consideration, but it is not obvious what the intervention should be.
Alternate means of communication in crisis
In communications-denied environments (where cellular networks and brittle hotlines fail), text and speech are not the only ways of communicating. One objection often mentioned is that leaders would communicate through their actions — tacit bargaining, as explained above. As Ash Carter explained, “The leaders of the United States and the Soviet Union would of course communicate through the violent actions they ordered.” Communicating de-escalatory intent may be very difficult in the fog of war, however. There may be other means of communication that do not rely on expensive hotline systems. To take an extreme example, one could even communicate via Morse code by exploding weapons — no networks required. Nonetheless, such communication would (1) be very costly, (2) be difficult to attribute to one’s adversary, as opposed to another state or rogue actor, and (3) be likely to be misinterpreted as aggression. One could also plausibly use autonomous drones to communicate, but again these would likely be misinterpreted as aggression (and funding work on this may accelerate harmful technological development).
Absence of Evidence
As explained in the first part of this shallow investigation, there is high uncertainty around hotline effectiveness for crisis management. The key difference between the crisis management scenario (application 1) and the war limitation and termination scenarios (applications 2 and 3), however, is the absence of other options for communication. In peacetime, it is unclear whether a hotline adds much value above other diplomatic channels (and importantly, the most crucial dyads — like U.S.-China — are already covered). In an escalating situation, however, the upsides of a resilient hotline may be larger if there are no other communications channels — a resilient hotline may be the only way of communicating de-escalatory intent (and here, the existing dyadic hotlines may break down). As explained above, even a tiny chance of de-escalation (1-in-1,000 nuclear conflicts) could have immense value.
A fictional example may illustrate the point: consider a nuclear war between India and Pakistan, which degrades normal communications channels, including communications between the U.S., Russia, and China (e.g., because space systems are down). Eager to limit the conflict to the subcontinent, and avoid all-out war between the superpowers, all three parties wish to communicate with each other and avoid misunderstandings and accidents (akin to the situation with the Six Day War, discussed above), but find that their “traditional” hotlines don’t work. Having the option of a resilient back-up hotline could be immensely useful in this scenario to avoid an even worse global catastrophe.
It is important to remember, however, that this is a hypothetical scenario, and we do not have a thorough understanding of hotline resilience in part because technical details of NC3-related resilience questions are classified. Moreover, we are even uncertain about the sign of hotline effectiveness — does it reduce or increase the risk of escalation? From a common-sense perspective, we may expect it to reduce the risk, but without more rigorous evidence, we cannot draw this conclusion.
What Could a Philanthropist Do?
Philanthropists could take a number of actions to try to address the problem of hotline resilience, and create a fail-safe communications mechanism between states:
- Fund work and dialogues to establish new hotlines;
- E.g., paying for track II dialogues between Israel and Pakistan to discuss the potential establishment of a hotline.
- Fund work and dialogues on hotline resilience (including technical work on hotlines in communications-denied environments);
- Fund more rigorous studies of hotline effectiveness;
- E.g., paying historians to attempt to quantify hotline effectiveness, or funding experimental wargames on hotlines.
- Fund track II dialogues between the U.S. and China (and potentially other powerful states) focused on hotlines to understand different conceptions of crisis communication or funding Chinese studies of nuclear risk reduction.
Establishing New Hotlines
As explained above, we think that funding work to establish new hotlines should be a lower priority. We can think of this in terms of a “multiplier” framework of effectiveness, where the highest-value dyads (those where war would be most destructive) have already been connected. This is especially true because of potential hotline lock-in. Current systems may be brittle in wartime, and spreading these systems further may simply spread hotline brittleness. Effective philanthropists could consider the most effective ways to sequence their interventions, and understanding the institutional, political, and technical issues behind better-functioning hotlines can be a first step towards implementing or updating new resilient hotlines all around the world.
Working on Hotline Resilience
A second potential intervention is to fund direct private sector work on the problem of hotline resilience. One project we are aware of in this space is CATALINK, a project of the Institute for Security and Technology (IST). We are not entirely without reference classes on the tractability of a project like CATALINK. First, we can construct a base rate on the frequency of hotline updates. As far as we are aware, since its establishment in 1963, the U.S.-Soviet/Russia hotline has been updated three times: once in 1971, once in 1984, and once in 2008. This suggests a naive base rate of 3 updates in 59 years, or an annual probability of hotline update around 5%. (This base rate is only for publicly-known full-system technical updates — other changes to the hotline were made, such as the 1967 move of the hotline from the Pentagon to the White House, or the choice to place clocks showing the time at the other end of the hotline.)
Second, we can find base rates for the success of “technology-focused ambitious R&D projects.” This has been studied in the context of transformative research at the National Science Foundation (NSF). The NSF’s Small Grants Exploratory Research Program focused on “high-risk, high-reward research that might not pass the traditional peer review process” and was evaluated in a 2013 study by SRI International. The study suggested that about 10% of grants resulted in the desired highly transformative “spectacular success” — along the lines of creating an entirely new field of research, like DNA-based computing. (A long-form discussion of DARPA — another analogous high-risk tech-focused program — by Benjamin Reinhardt claims a 5-10% success rate for DARPA projects, but gives no evidence for this claim.) While imperfect, this can function as a rough order-of-magnitude success rate for highly speculative STEM research.
Open Questions for Technical Hotline Projects
There are several open questions for technical hotline projects. First, do hotlines matter? If they matter, do they do more good than bad in a crisis? There is little evidence on this, but the question has potentially large consequences. If hotlines were completely counterproductive, philanthropists and governments could save many millions of dollars for other risk-reduction projects. If hotlines were highly effective, on the other hand, and highly brittle, hotline resilience projects may be among the most tractable longtermist risk-reduction measures available.
Second, can we address the political and institutional problems that appear to prevent effective functioning of hotlines even in peacetime? As we have explained elsewhere, we view U.S.-China competition as one of the most concerning international relationships of our time, with implications not only for nuclear war but also for emerging technologies and semiconductor supply chains. This dyad, however, is also the most concerning hotline relationship, in part because of China’s repeated failure to “pick up” in times of crisis.
Finally, what are the necessary technical specifications of a crisis communication system to be viewed as “trustworthy” by states, and what are the real technical vulnerabilities of hotlines between the most powerful states? This question may be unanswerable in an unclassified forum, such that the best course of action is for private philanthropists to attempt to stimulate government action and catalyze necessary technical updates without “guessing” at what those updates might be.
Funding Track II Exchanges on Crisis Communication
As mentioned above, before working on specific technical problems, we ought to consider the political and institutional obstacles to better crisis communication. For example, why did China not pick up during the Hainan Island incident? Does the leadership of the Chinese Communist Party view crisis communication in the same way that U.S. leadership appears to view it (i.e., as a risk-reducing confidence-building measure), or does it view these systems as a mechanism for gaining the upper hand in a crisis situation? Crucially, are states aware of the problem of hotline resilience, if it is a real problem?
Thus, one potential intervention is to fund track II dialogues between powerful states to better understand different views of crisis communication (e.g., between the U.S. and China). This intervention would operate on multiple levels. First, as a track II dialogue, it can operate as a confidence-building measure per se, regardless of content, that helps to build greater trust and understanding between the powers. Second, it could have high value as a “trojan horse” for risk reduction — a conversation about hotlines superficially appears as a conversation about diplomatic processes, but can also function as a way to discuss sensitive issues like NC3, space assets, and cyber vulnerabilities, which may otherwise not be discussed. Third, it could have high value in pointing to whether further private sector projects are needed. Finally, it could have high value for states in understanding the vulnerabilities of their own systems, and raising awareness of the risks of hotline failure.
Track II exchanges can be relatively cheap, with most of the expenses consisting of travel and lodging for participants. Importantly, this could catalyze government funding. If a track II dialogue were to reveal a previously unexamined weakness in NC3 systems, governments could invest billions to correct these weaknesses.
Studying Hotline Effectiveness
Funding more rigorous studies of hotlines could be a valuable intervention, given the high uncertainty around their effectiveness. There are some problems with historical studies, given issues of incomplete and selective declassification, and given the problem of counterfactuals. Hotline related knowledge-production could be more tractable on quasi-experimental methods, simulations, and wargames. Studies of hotline effectiveness could have two aims:
- Producing better knowledge about hotline effectiveness;
- Raising government awareness of hotline resilience problems.
The first effect, producing better knowledge, has informational value for both governments and philanthropists, who can let this knowledge inform their decision-making and funding. The second effect, raising awareness of hotline problems, could be especially valuable in catalyzing government action. If governments are simply unaware of the problem of hotline resilience, then reading about (or even participating in) a wargame on this problem could help stimulate action to build more resilient systems.
The term “wargame” may connote specific visions of table-top exercises or Cold War musing about outcomes. We are using the word in a more expansive way that is closer to a subset of psychology experiments, as defined by Lin-Greenberg et al.: “interactive events that display four characteristics: human players, immersed in scenarios, bounded by rules, and motivated by consequence-based outcomes.”
For example, a philanthropist could fund the following experiment on hotline failure. After selecting participants (either a sample of elite decision-makers or a more random sample of the general population), the investigators could randomly assign them to one of two scenarios:
- A nuclear crisis where teams have access to a communications channel (a “hotline”) to explicitly communicate and bargain with other “states.”
- A nuclear crisis where this communications channel has failed and participants rely on tacit bargaining to communicate with other “states”.
If appropriately designed and repeated, variations in outcomes between the two scenarios may point towards the effect of the absence/presence of a crisis communications channel. If participants from other countries can be recruited, the wargame can simultaneously function as a type of track II dialogue — participants from the U.S. and China, for example, might gain new insights about how the other side views such crises and the role of hotlines in managing them.
This kind of experiment is relatively new in international relations research. Most fundamental is the question of “ecological validity,” or “the extent to which behavior under test conditions mirrors real-world behavior.” In other words, what can human behavior in a simulated crisis tell us about human behavior in a real crisis? This objection, however, is not as strong as it appears. First, the experimental environment can be designed to be as immersive and close to the “natural” environment as possible. Experimenters can also conduct surveys after the wargame to gauge whether the scenario “felt real” to participants. More broadly, however, the issue of generalizability affects all science, especially studies of human behavior. Even historical evidence about “real” crises have limited value, and it is not clear whether a full study of U.S.-Soviet crisis communication would generalize well to other contexts in the present day. In a world with (fortunately) limited information about global catastrophic risks, experimental wargames — which can benefit from randomization, controlled variation in the experimental variable, and large sample sizes — may be among the best flawed tools available. If such an experiment were to find a repeated large effect, for example, it could open up further qualitative research to understand the causal mechanism behind that effect. Thus, wargames may be among the best tools we have to study the unprecedented event of nuclear war, as Herman Kahn pointed out when military officers criticized the Cold War strategists’ lack of “real world” experience: “How many thermonuclear wars have you fought recently?”
Finally, ecological validity is not necessary for raising government awareness. U.S. decision-makers have long used wargames, including poorly-designed non-experimental wargames, to inform defense policy since the early Cold War. Whether or not a wargame produces valid knowledge, therefore, it can nonetheless stimulate government action on important issues. For example, a high-ranking participant in a wargame may not act as they might in a real crisis, but they may learn that there is a blindspot in hotline resilience that they had not previously considered, and which they can correct.
Importantly, wargames are relatively cheap. After discussing the issue with one expert on wargames, we estimate that a single game with about 100-200 participants can cost around $25,000 to $50,000 (much less if the game includes only local participants, more if it includes much international travel). For the cost of a large technical project ($10 million) on hotline resilience, therefore, a philanthropist could fund a major scientific project of 200 high-quality wargames.
Catalyzing Government Action
We have mentioned the issue of catalyzing government action several times in this report, and believe that there are strong reasons to prefer leveraging government funds over funding a large private-sector project.
Beyond Nuclear War
Hotlines could be a useful tool for communicating on a variety of catastrophic risks, including nuclear incidents, AI-related incidents, incidents with synthetic biology or laboratory leaks, etc.
For example, CATALINK is explicitly designed with non-nuclear use cases in mind: it could be deployed “within the national NC3 infrastructure as well as for multi-party communications in other non-nuclear but catastrophic situations, such as natural disasters, global pandemics, nuclear-plant meltdown, or other circumstances where existing communications systems may be compromised or unavailable.” This means that the above considerations may under-estimate the value of such a project.
Moreover, even if such a project is not adopted by states, the technology developed as part of this project could be helpful for emergency response teams. As CATALINK investigators write, “a communications system that can work under high-stress conditions could be valuable for an enterprise incident-response team whose infrastructure and credentials have been compromised.”
The Funding Landscape
Nuclear command and control is well-funded within governments, but inter-state crisis communications appears to be neglected. Overall, the U.S. spends around $4 billion a year on operation, maintenance, and upgrading of NC3 systems. Digging deeper into the budget for advanced technology development, however, only $5,098,000 were requested for “NC3 Advanced Concepts.” From the 2018 Nuclear Posture Review, moreover, it appears that crisis communication between states is not counted as part of NC3 at all; the NPR lists the “five major functions” of NC3 as “detection, warning, and attack characterization; adaptive nuclear planning; decision-making conferencing; receiving Presidential orders; and enabling the management and direction of forces.”
Despite the large-scale effort to modernize NC3, we were unable to find any evidence of governmental efforts to make inter-state hotlines like the Washington-Moscow hotline more resilient. There are several potential explanations for this. First, absence of evidence is not evidence of absence, and there may be efforts that are classified or low-profile. Second, nuclear powers have a poor track record at taking obvious steps to make nuclear hotlines resilient. This is perhaps illustrated best by the Finnish farmers’ plowing through the original hotline cable — an easily-anticipated disruption. Third, while states have incentives to work on their own NC3 systems, tense inter-state relations (like U.S.-China competition) make cooperation difficult without help from private or neutral parties (as illustrated by track II dialogues). Finally, direct state participation on hotline resilience may require acknowledging deep vulnerabilities in one’s NC3 systems in general, something that all states would be reluctant to do for fear of undermining their deterrent’s credibility.
Even in the private sector, hotlines are relatively neglected. Searching the Candid Peace and Security funding database for hotline-related keywords — “hotline,” “DCL,” “MOLINK,” “Direct Communications Link,” “crisis communication” — yielded no results. Indeed, since 2012, Candid lists only the following grants even tangentially related to nuclear communications:
“For a project on emerging Anti-satellite (ASAT) threats to U.S. Nuclear Command, Control and Communications (C3) space assets”
2017 Carnegie Corporation: $450,000 to Carnegie Endowment for International Peace “For a project on cyber threats to nuclear command and control systems”
|"For a project on cyber threats to nuclear command and control systems"|
|“For investigating disruptive technologies; strategic vulnerability; and the future of deterrence. Additional key words provided by funder: proliferation; command and control; cruise missiles”|
|“For assessing the impact of disruptive technologies on strategic stability between the United States and Russia. Additional key words provided by funder: U.S.-Russia; nuclear issues; command and control; space”|
|“For a scenario-based project on the dynamics of command; control; and coordination in cyber-conflict escalation. Additional key words provided by funder: command and control”|
Technology for Global Security
|"X-grant for a project that addresses technical vulnerabilities of emerging nuclear command, control, and communications."|
|"in support of a project to develop an international nuclear command, control and communications code of conduct."|
In light of MacArthur’s departure from the nuclear space, this area may be in special need for funding.
The theory of change for hotlines is clear — in some situations, their existence could save billions of people, and positively affect the long-run trajectory of humanity. There is high uncertainty about hotline effectiveness in crisis management, however, and no evidence on hotline effectiveness in war termination or limitation (because there has been no nuclear war). The existence of key hotlines — especially between great powers — and the apparent redundancy of rapid communications links together make the establishment of new hotlines a low priority. NC3 modernization itself is not neglected, thanks to DoD spending, but hotline resilience appears to be highly neglected, both in government and the private sector.
Several key questions remain unanswered:
- How large is the potential effect of hotlines in reducing the probability escalation?
- How important is hotline resilience (and could uncertainty about resilience be beneficial itself)?
- How can U.S.-China crisis communication be improved?
It is possible that hotlines have high value in reducing global catastrophic risk, but it is also possible that they are useless or even counterproductive. I recommend research and dialogues on hotline effectiveness and U.S.-China crisis communications as next steps in understanding hotlines as a potential intervention.
About Founders Pledge
Founders Pledge is a community of over 1,700 tech entrepreneurs finding and funding solutions to the world’s most pressing problems. Through cutting-edge research, world-class advice, and end-to-end giving infrastructure, we empower members to maximize their philanthropic impact by pledging a meaningful portion of their proceeds to charitable causes. Since 2015, our members have pledged over $8 billion and donated more than $800 million globally. As a nonprofit, we are grateful to be community-supported. Together, we are committed to doing immense good. founderspledge.com
Egillson, “The Origins, Use, and Development of Hotline Diplomacy,” 1-2.
“Hot Line Agreement,” U.S. Department of State, accessed August 17, 2022, https://2009-2017.state.gov/t/isn/4785.htm
Desmond Ball, “Improving Communications Links between Moscow and Washington,” Journal of Peace Research 28, no. 2 (May 1991): 135, https://doi.org/10.1177/0022343391028002002.
Egillson, “The Origins, Use, and Development of Hotline Diplomacy,” 2-3, quoting Dobrynin, In Confidence: Moscow’s Ambassador to America’s Six Cold War Presidents, 96.
Real-time voice communication and translation was seen as too error prone and more likely to lead to heated exchanges. Apparently, the common misconception that the link is in fact a “red telephone” is not limited to the public, and even officials of the U.S. State Department appeared to believe in 1983 that the hotline included speech capability. (See Egillson, “Origins,” 5.)
Desmond Ball, “Improving Communications Links between Moscow and Washington,” Journal of Peace Research 28, no. 2 (May 1991): 136, https://doi.org/10.1177/0022343391028002002. Steven E. Miller, “Nuclear Hotlines: Origins, Evolution, Applications,” Journal for Peace and Nuclear Disarmament 4, no. sup1 (March 5, 2021): 182, https://doi.org/10.1080/25751654.2021.1903763.
Note, however, that these vary on important dimensions. Not all are used as crisis communications channels, and not all connect heads of state. Whether all of these communications links ought to be called “hotlines” is therefore debatable.
“Japan and China Agree to Set up Defense Hotline amid Tensions over East China Sea, Taiwan - CNN,” accessed August 29, 2022, https://www.cnn.com/2021/12/28/asia/japan-china-military-hotline-intl-hnk/index.html.
Note, however, that there are situations where crisis communications between allies may be especially valuable, as when a great power keeps a smaller ally “in line” to prevent escalation of a regional conflict.
Desmond Ball, “Improving Communications Links between Moscow and Washington,” Journal of Peace Research 28, no. 2 (May 1991): 135, https://doi.org/10.1177/0022343391028002002.
“The strengthening or building of confidence between governments has been one of the most valuable aspects of hotlines. They have increasingly been used as an important first step in building or enhancing confidence between both friendly and unfriendly states, symbolizing improved relations or intentions to do so.” (Egillson, “The Origins, Use, and Development of Hotline Diplomacy,” 20.)
Schelling, “Bargaining, Communication, and Limited War,” Journal of Conflict Resolution (reprinted), https://lps.library.cmu.edu/NCMR/article/87/galley/90/view/.
Weinberger, 1983, quoted in Ball, “Improving Communications Links,” 138.
Arms Control Association, “Hotline Agreements,” accessed August 8, 2022, https://www.armscontrol.org/factsheets/Hotlines.
Arnav Manchanda, “When Truth Is Stranger than Fiction: The Able Archer Incident,” Cold War History 9, no. 1 (February 2009): 125, https://doi.org/10.1080/14682740802490315.
Haraldus Þór Egilsson, “The Origins, Use and Development of Hotline Diplomacy,” 29.
Thanks to Peter Rautenbach for pointing to this in a round of external review.
Lyle J. Morris and Kyle Marcrum, “Another ‘Hotline’ with China Isn’t the Answer,” July 27, 2022, https://www.rand.org/blog/2022/07/another-hotline-with-china-isnt-the-answer.html.
Shirley A Kan et al., “China-U.S. Aircraft Collision Incident of April 2001: Assessments and Policy Implications,” https://sgp.fas.org/crs/row/RL30946.pdf
Seymour Hersch, “The Online Threat,” The New Yorker, accessed August 18, 2022, https://www.newyorker.com/magazine/2010/11/01/the-online-threat.
Richard Armitage, quoted in Shirley A Kan et al., “China-U.S. Aircraft Collision Incident of April 2001: Assessments and Policy Implications,” 13, https://sgp.fas.org/crs/row/RL30946.pdf
Julian Borger, “Hotlines ‘Ring out’: China’s Military Crisis Strategy Needs Rethink, Says Biden Asia Chief,” The Guardian, May 6, 2021, sec. World news, https://www.theguardian.com/world/2021/may/06/hotlines-ring-out-chinas-military-crisis-strategy-needs-rethink-says-biden-asia-chief-kurt-campbell.
Jack Detsch, “Pentagon Hopes for More China Hotlines,” Foreign Policy, accessed August 29, 2022, https://foreignpolicy.com/2022/06/09/pentagon-china-hotlines-diplomacy/.
William MacAskill, What We Owe the Future, 32.
Ball, “Improving Communications Links,” 157.
Ball’s own inclusion of the case appears to have been based on contemporary reporting by a British tabloid, decreasing my confidence in this case as evidence for hotline effectiveness.
Sudhi Ranjan Sen and Faseeh Mangi, “Errant Indian Missile Nearly Led to Pakistan Retaliatory Strike,” Bloomberg, https://www.bloomberg.com/news/articles/2022-03-15/errant-indian-missile-nearly-led-to-pakistan-retaliatory-strike?leadSource=uverify%20wall.
“Call Me Any Time: Trump Says He Gave North Korea’s Kim Direct Number,” Reuters, June 15, 2018, sec. APAC, https://www.reuters.com/article/us-northkorea-usa-trump-idUSKBN1JB1R5.
Schlesinger, quoted in Ball, “Improving Communications Links,” 139.
Ashton Carter, 1987, quoted in Last Chance, 4.
Ball, “Improving Communications Links,” 139.
“Today NC3 systems are in fact ‘systems of systems’ that rely on legacy and modern technologies that are increasingly vulnerable to digital and other rapidly emerging, disruptive capabilities.” (Last Chance, 2). For more on the resilience of defense systems, see Defense Science Board, Resilient Military Systems and the
Advanced Cyber Threat.
“Work Underway for Next Generation Nuclear Command Control and Communications,” accessed August 17, 2022, https://www.nationaldefensemagazine.org/articles/2021/1/5/work-underway-for-next-generation-nuclear-command-control-and-communications.
“This patchwork character leads many experts to believe that these systems will not be resilient enough to overcome new threats arising from cyberwarfare, autonomous vehicles, and artificial intelligence—technologies that could dramatically accelerate the escalation of nuclear-prone conflicts and add uncertainty at the exact moments when commanders must make decisions with existential import” (Last Chance, 5)
Ball, “Improving Communications Links,” 142.
Ball, “Improving Communications Links,” 137.
Last Chance, 30.
“Systems for communications, indications and warning, and response capabilities are likely to be among the
first attacked early in a crisis.” (Last Chance, 5).
“Today’s systems comprise elements from myriad sources, many of which could plausibly be malicious. Many elements are also so complex that verification of the absence of malice is difficult or impossible to achieve. We should understand this intuitively, as even world-class companies regularly experience failure or compromise.” (Last Chance, 6)
Ball, “Improving Communications Links,” 139.
Ball, “Improving Communications Links,” 143.
Last Chance, 9.
Last Chance, 9.
Last Chance, 9.
Last Chance, 13.
Future of Life, Accidental Nuclear War: A Timeline of Close Calls, https://futureoflife.org/background/nuclear-close-calls-a-timeline/.
Importantly, we do not mean to minimize the destruction and suffering of any kind of nuclear war, but simply point out that this destruction and suffering would occur many times worse in some nuclear wars than in others.
These estimates are for illustrative purposes only. Rodriguez (2019), gave a 90% confidence interval of 2.7 billion to 7.5 billion deaths in the event of a nuclear winter triggered by all-out U.S.-Russia War (Rodriguez, “How bad would nuclear winter caused by a US-Russia nuclear exchange be?”)
This is an old idea. See, e.g., Thomas Schelling “It is not invariable an advantage, in the face of a threat, to have a communication system in good order, to have complete information, or to be in full command of one’s own actions or of one’s own assets.” (Schelling, Strategy of Conflict, 18)
Ash Carter: “The leaders of the United States and the Soviet Union would of course communicate through the violent actions they ordered.” quoted in Last Chance, 6.
It is possible that there have been hotline updates that were not publicly disclosed, e.g. because they involve classified technology.
Notably, these are all bilateral updates, and do not include the probability that either the U.S. or Russia may take unilateral steps to harden or secure their crisis communications systems.
Haraldus Þór Egilsson, “The Origins, Use and Development of Hotline Diplomacy,” 8-9.
See Founders Pledge’s Great Power Conflict report for an in-depth discussion of track II dialogues.
Lin-Greenberg et al., “Wargaming for International Relations Research,” 5.
Lin Greenberg et al., “Wargaming for International Relations Research,” 8.
Kahn, quoted in Sharon Ghamari-Tabrizi, “Simulating the Unthinkable: Gaming Future War in the 1950s and 1960s,” 165.
Sharon Ghamari-Tabrizi, “Simulating the Unthinkable: Gaming Future War in the 1950s and 1960s.”
Last Chance, 8.
Last Chance, 8.
John R Hoehn, “Nuclear Command, Control, and Communications (NC3) Modernization,” Congressional Research Service.
Department of Defense FY 2023 Budget, “RDT&E Programs,” https://comptroller.defense.gov/Portals/45/Documents/defbudget/FY2023/FY2023_r1.pdf.
U.S. Department of Defense, 2018 Nuclear Posture Review, https://media.defense.gov/2018/Feb/02/2001872886/-1/-1/1/2018-NUCLEAR-POSTURE-REVIEW-FINAL-REPORT.PDF.
Descriptions from Peace and Security Funding Index.