Since the Cold War era, submarines carrying nuclear weapons have quietly hidden within the depths of the ocean, playing an important role in the delicate art of nuclear deterrence. The logic of using submarines with nuclear weapons capabilities as undersea deterrents goes like this: if you make it known that you have nuclear warheads concealed at sea, your enemy knows you will always have a means to strike back. This undersea presence is meant to serve as a disincentive for first strike nuclear attacks.
But this logic only works if you can keep your submarine hidden.
In 2020, nuclear powers (and their allies) find themselves engaged in a more complex game of deterrence—particularly under the sea. There are approximately nine nuclear-armed states today, six of which are known to maintain or are believed to be planning to deploy submarine-based nuclear weapons programs. Technology is improving, and countries are investing – both in enhanced submarines and in better methods for detecting and disabling them. This changing landscape begs the question: Will submarines with nuclear capabilities remain a relevant tool for nuclear deterrence?
There is no certain answer to this question, but our technological history and futures play a role in whether and how the seas remain a “safe” harbour for nuclear weapons. In the arms race of ‘hide and seek’ it is disruptive technologies, such as ocean sensing, data analytics and Artificial Intelligence that may forever change the balance between the hider and seeker and in so doing complicate and even disrupt the already tense social, political, and strategic dynamics of nuclear deterrence.
With this in mind, three 3Ai staff members — Katherine Daniell, Ehsan Nabavi and I — recently took part in a multi-disciplinary collaboration exploring these possible futures of undersea deterrence from a scientific and technological lens. This work, led by ANU Professor Roger Bradbury, was designed to investigate whether the very premise of maintaining a stealth nuclear capability would remain viable in the face of advances contributing to advanced underwater detection and counter-detection methods over the next 30 years.
The central question behind our study was this: Will future science and technology make the oceans transparent?
Our findings, which were recently published by the ANU National Security College, suggest that the future of undersea deterrence is likely to look very different. In particular, the advantages nuclear-capable submarines (SSBNs) currently leverage in deterring a nuclear attack are likely to evaporate.
The term “likely” is important here – one that has specific meaning in the context of our study. We are not fortune tellers, and scientific or technological progress is never linear. The ocean is also not homogenous: we may use a single word to describe a vast collection of seas with diverse geographies, populations, and interests, but what becomes true for one region may not make sense for another. What we can do, however, is identify areas of science and technology development that are likely to be relevant to our study’s question and make judgments about whether relevant advancements are likely to occur in the time frame we’ve specified. We are, in other words, looking for causal relationships between technological developments and outcomes of relevance to our problem, and assigning probabilities (75% in the case of “likely”) that a technological development will result in a causally linked outcome.
In other words, we create a model—one which we can then use to predict an answer to our question. The approach we chose used subjective logic – an approach that is particularly good for uncertain problems like ours.
Those involved in crafting the model decide on the model structure: a hierarchy of sub-questions whose possible answers (structured in the form of opposing outcomes) may influence the answer to our main research question. This is an inherently subjective process, shaped by the perspectives of those involved (much like most models in existence—even the most rigorous began with a string of hypotheses crafted by human intuition). Diversity is therefore essential for this—and I would argue, any—modelling process. With this in mind, the team involved in shaping the model had a wide range of backgrounds, covering fields ranging from water policy, marine science, artificial intelligence, nuclear physics, engineering, national security, and complex systems science.
Once we shaped the model, we then sought evidence we could use to explore the model’s question hierarchy. This, too, was a subjective process, but again, this is explicitly taken into account within the model. How trustworthy is a given piece of evidence? And what does this evidence suggest about the likely answers to our hierarchy of questions?
The model—as we created it—is presented in detail within our publication. And the answer it gives to our question? The oceans are likely—and even very likely, in some respects—to be transparent by the 2050s.
The model we developed is inherently subjective, and we expect—as time goes on, and as technology develops—to need to add to its structure, tweak its probabilities, and supplement (or even replace) the evidence we used to come to an answer. These modifications and additions may change our assessment, and our conclusions. From my perspective, though, this is one of the model’s great strengths: its structure is easily interrogable and modifiable—adaptive in the face of new evidence.
Does this tell us the future of nuclear deterrence over the next 30 years? Perhaps not. But it gives us an incentive to continue asking questions about the future of deterrence. For 3Ai, there are questions about how the technological and human systems that evolve to both detect and hide SSBNs in the future might intentionally or otherwise shape our global futures.
How will this system of undersea technologies interact in an increasingly crowded ocean? How will they signal their intentions, and how will those intentions be perceived? What actions will result?
None of these questions have easy answers—and all, in some sense, require an understanding of the many social, cultural, political, and environmental factors shaping the designs of any system involved in nuclear deterrence.
Transparent Oceans? The Coming SSBN Counter-Detection Task May Be Insuperable Authors: Roger Bradbury, Scott Bainbridge, Katherine Daniell, Anne-Marie Grisogono, Ehsan Nabavi, Andrew Stuchbery, Thomas Vacca, Scott Vella and Elizabeth Williams.