North Korea’s nuclear test on September 3 was registered as a rare literal geopolitical earthquake. Some public uncertainty persists about the scale and significance of the tremor. It has been reported in a range of magnitudes from 6.1 to 6.3 (or even higher), on the logarithmic Richter Scale. An event of this size suggests an explosion of several hundred kilotons of TNT, and is consistent with the detonation of a thermonuclear device. North Korean confirmation of exactly this occurrence has been received with unprecedented seriousness.

Nuclear non-proliferation is more idea than reality. Its only substance is a comparative sluggishness when estimated against the benchmark of some generally unstated nightmare scenario. According to such counter-factual consideration, nuclear weapons might have been more widespread than they are by now. But exponential processes look like this. They start small, and don’t seem to be going anywhere dramatic for a while. As the celebrated fable of exponentiation shows, a modest bowl of rice gets you quite a long way into the chess board. The supposedly common-sense assumption that uncontrollable nuclear proliferation isn’t yet happening requires an argument. (This short essay makes the other argument.)

The nuclear ‘club’ is too unwieldy to share any kind of seriously constraining principle. There is nothing identifiable that entitles a nation to membership, beyond naked possession of doomsday-tier military capability. The club was trans-ideological from the start, and quite soon afterwards highly multicultural. Among members, reciprocal distrust and even hostility is the norm, which – given the runaway action-reaction process that settled the membership roster – could scarcely be unexpected. The behavior of members is controlled by nothing beyond game theory. It’s also very much worth mentioning that nobody who manages to get into the club can, in any practical way, be thrown out.

The United States detonated the world’s first thermonuclear, two-stage, fusion, or (Teller-Ulam design) ‘hydrogen’ bomb at Enewetak Atoll on November 1, 1952. The Soviet Union responded less than a year later, testing its own H-bomb on August 12, 1953. Tests – or demonstrations – followed in succession from The United Kingdom (November 1957), China (June 17, 1967), and France (August 1968). Israel is thought to have conducted a joint test with the Republic of South Africa – the so-called ‘Vela Incident’ – in September 22, 1979. In 1991 the South African government claimed to have assembled, and later unilaterally dismantled, six nuclear devices. India expanded the spiral of thermonuclear proliferation into South Asia with a test in May 1998. Pakistan is not known to have tested anything beyond ‘boosted fission’ devices, but it formidable nuclear capability is not in question. (A longer essay would have found space at this point to acknowledge Pakistani Abdul Qadeer Khan’s disproportionate contribution to the global proliferation dynamic.) Saudi nuclear cooperation with Pakistan can be expected to accelerate the spread of nuclear weaponry into the Arabian Peninsula, once Iranian progress in the military application of the technology triggers the long-anticipated Sunni-Shia arms-race in weapons of mass destruction. Hence the chain of proliferation steadily lengthens on its main axis, through Cold War superpower rivalry, into Chinese triangulation, a responsive Indian bomb, and then into the fractured world of Islam, via Pakistan (with unreciprocated Israeli nuclear prowess as additional prompt, and pretext).

The one-dimensional character of this narrative is an artifact of its immaturity. The under-development of the proliferation process appears to present the ‘international community’ with no more than a single crisis at any time. Things will not look this way for long. There is nothing essentially mono-linear about the dynamic of cross-escalation. Increasing momentum is already taking it off the tracks. As Richard Fernandez notes, lines of nuclear escape are occurring in several directions at once:

In security affairs the old East-West game payoff matrix has been replaced by a multidimensional array of new players many of them sub-national, some of them unknown. The big wild card is technology. Disruptive technological change and new modes of warfare associated with them have upset the old calculus. North Korea, Iran are not outlier threats but leading indicators of the changed dynamic. They are the first samples of a new threat coming onstream.

North Korea claims to have tested thermonuclear weapons in January 2016, following fission device tests in 2006, 2009, and 2013). Whether as a matter of analytical realism, or of strategically motivated public skepticism, the claim was met by orchestrated Western disparagement. The 2017 test shattered this wall of denial. In the words of Scott D. Sagan, writing at Foreign Affairs: “North Korea no longer poses a nonproliferation problem; it poses a nuclear deterrence problem”

While, if traced as a simple historically consistent curve, it is not yet impossible to see a process of deceleration in this time-line, such an optic is ceasing to convince. It seems to be part of a collapsing world order, which is taking its structures of perception down with it. The assumption of continuity, for instance, now seems reckless in the extreme. Historical discontinuity in the proliferation dynamic has been especially notable over recent decades, due to a hardening pattern whose incentive effects could not easily be more ominous. The surrender of thermonuclear ambitions has acquired a stark correlation with subsequent regime destruction, unlike anything seen in the previous era of Cold War superpower patronage.

Ukraine voluntarily surrendered its nuclear arsenal to Russia upon the disintegration of the Soviet Union. In the Gorbachev era, this decision no doubt appeared rational – and even prudent. Subsequent regional developments make it far harder to excuse. It remains to be seen whether Ukrainian national independence will have finally been sacrificed to this high-minded call, but rudimentary geopolitical and domestic security already has been.

The prevailing racial hysteria of our age hazes any analysis of South African regime change in comparable terms, as it already hazed the process itself. Future historians will have clearer eyes. It certainly seems to fit the pattern. No less than with Juche, the experience of apartheid is that sensitivity to international ‘polite opinion’ is vastly increased by the absence of nukes.

The Libyan lesson has been the most lurid to date. Libyan denuclearization “was peacefully resolved on December 2003” Wikipedia explains. In a separate article it adds the appendix (more helpfully still) that “Muammar Gaddafi, the deposed leader of Libya, was captured and killed on 20 October 2011 during the Battle of Sirte. … videos of his last moments show rebel fighters beating him and one of them sodomizing him with a bayonet before he was shot several times as he shouted for his life.” It would be difficult to devise a more graphic educational resource against international WMD non-proliferation compliance.

This is the background against which North Korean nuclear obstreperousness is to be gauged. The regime had, in any case, already made obnoxiousness into a local specialism. Its delinquent international behavior has long been the stuff of dark comedy. The country’s cultivated image takes prickly into territory the zoological porcupine lineage has yet to explore.

In respect to strategic fundamentals, however, the regime’s feral punk-performance attitude to diplomatic conduct is not the principal issue. Bad attitude makes for stimulating diplomatic theater, but it decorates the fundamentals of threat. Focus on capabilities, not motivations, is a strategic principle that cannot be over-stressed. In the case of North Korea, and others no doubt soon to follow, however, it is a principle that requires complete inversion. A definite incapacity rises, instead, to strategic prominence.

The extremity of the emerging North Korean threat is a function of weakness, in many respects, but most centrally regarding its new responsibilities for deterrence management. Insecure nuclear arsenals are destabilizing, since they incline to first use, on the use-it-or-lose-it principle. Vulnerability to a first-strike is a continuous prompt to pre-emption.

North Korea is a geographically small nation, with crude command-control structures, very limited early warning capabilities, and an exclusive reliance on exposed land-based ballistic missile platforms for warhead delivery. In other words, it is destined to remain on a hair-trigger from the moment it crosses the deterrence threshold. Rather than being a splitting headache to the world order by relentless, malignant initiative, it will henceforth be one by simple strategic default. The world will have become a city built under Vesuvius, quite regardless of any planning decisions or philosophies of risk. An epoch of peril is opening.

Under these conditions, mere ‘capability’ becomes extraordinarily provocative, and incompetence is automatically terrorizing. Yet, while this dilemma is not difficult to understand, it is perhaps a little too difficult to be captured by any public debate conducted at a realistically imaginable level of sophistication. Insofar as there is anything like a court of global mass opinion, it can be confidently expected to miss the strategic essentials and lose itself in multilateral theater performances. Geostrategic realities and mass perceptions are on diverging trajectories.

The prevalent delusions tend to be simplifying, and retarded (in the strict sense). They lag the diffusive trend, and thus invoke unrealistically economical structures of agency, drawn back towards the long-lost ideal of bipolarity. The age of superpowers still dominates the nuclear imagination.

Because there is no road through Pyongyang that doesn’t end in a pit full of diplomatic punji sticks, the temptation is to fantasize a road through Beijing. No such thoroughfare exists. Relations between China and the North Korean regime have reached their lowest point since the Korean War, and are now frankly hostile. The Kim Jong-un regime has sought to extirpate Chinese influence from its leadership, with spectacular ruthlessness. Targeting of Chinese urban centers by the North Korean arsenal is no longer unimaginable, or, in China, unimagined. After all, the natural target of a deterrent is the greatest threat to the wielding nation’s sovereignty. It is near-inevitable that China will occupy this role in the North Korean case. Chinese impotence in respect to North Korea is what the North Korean nuclear arsenal is largely – and perhaps even primarily – about.

Tyler Cowen describes Robert Heinlein’s (1966) The Moon is a Harsh Mistress as “perhaps the best novel for understanding the logic of a future conflict with North Korea.” He then adds: “furthermore Catalonians should read it too. Most of all, I recall upon my reread that this book was my very first exposure to game-theoretic reasoning.” Not only exotic bombardment (by “electronic catapult”), independence struggle, and games, but also a world order reconstructed by the rise of China, and even a “malicious AI” who acquires strategic agency. Evidently, already half a century ago, Heinlein is exploring a durable cluster of concerns. At the very core: There can be no question of achieving or maintaining independence without the capacity to inflict serious harm upon those who might seek to prevent it.

Independence, in its geopolitical sense, fuses liberty and security indissociably. Autonomy – which is exactly sovereignty – requires insensitivity to coercion, and is thus the negative of foreign compelling threats. The analytical equivalence between reciprocal independence and a ‘balance of terror’ submits national autonomy to a geopolitical form of general relativity. Since no such thing as absolute security is realistic, sovereignty exists only in degrees, within tense networks. The tension is the game.

Thomas Schelling’s pioneering application of game theory to nuclear strategy remains the point of ingress into this world. The core reality of MAD games is easily misunderstood. Massive (or non-reiterating) retaliation is – at the stage it comes due – by immediate estimation irrational. It is then too late to contribute anything but compounded harm, regardless of its occasion. Under hypothetical conditions of amnesia and unconstrained action, it can never make sense. Yet, paradoxically, the ability to make credible retaliatory threats is the basic underpinning of rationality during prior negotiation games. Without it, there can be no reason for competitor restraint. The requirement, then, is for a future agent to be firmly committed to a conditional course of action that – at the potential point of execution – will be non-compelling.

Mutual assured destruction has been derided for its madness, but it is no less an outer-limit of sanity. Its logic is as rigorously implacable as any found within the social and historical sciences. The extreme moral disturbance that it arouses speaks in favor of its uncompromised rationality. Anguished intuition counts for nothing in its cold calculus, unless as a technical obstacle. The fact that people find this logic of inherited fatal commitments intolerable, as dramatized with exceptional vividness in the opening sequences of the 1983 movie WarGames, is our problem. The process is re-routed by our squeamishness, but not at all derailed. It has long been suspected that humans are too weak for MAD.

As an expression of absolute commitment, suicide terrorism appears to provide MAD with a microscopic model, but it is a weak and misleading one. Beyond difference in scale, suicide terrorism fails through execution. It communicates through actualization – or demonstration of will – which is the negative of deterrence. (Or perhaps, deterrence of a kind, expensively purchased.) The terror at the edge of the present, and of the future, has different models. Among these, civilization-scale ‘quantum suicide’ is perhaps the most exotic philosophical and ideological conception on its way to us. Given the assumption of a (Level-3 or higher) multiverse, comprehensive apocalypse is rationalized as the pruning of sub-optimal branches. It operates as reality editing. The game theoretic consequences of such a perspective are intriguing. It increases the credibility of threats (if accepted as a serious intellectual commitment), while adapting the pay-off matrix in a fashion that can only be considered destabilizing. Classic MAD works best among those who envisage an outcome as the worst thing in the world, yet commit to it anyway.

We approach here one of the very deepest problems in social and institutional engineering. It might be called the Odysseus Problem. In sailing past the Sirens, Odysseus anticipated the subversion of commitment, and thus put in place a socio-technical mechanism to bind his own future action. The structure is that of a ‘chicken game’ – a mutant variant of prisoner’s dilemma, in which the player who swerves loses. If you could back down, you might. In both Odysseus’ dilemma and that of the chicken player, the elimination of future discretion figures as a strategic resource. The requirement for self-binding inclines to a technological freezing of decision. Strategic problems of the ‘chicken game’ type thus tend inexorably to automation.

If AI is brought into play by the intrinsic dynamics of nuclear confrontation, it does not stop there. AI has a WMD potentiality proper to itself. There is no obvious horizon to what an algorithm could do. The same capabilities that enable algorithmic control of WMD arsenals equally enable such arsenals to be swapped-out for AI itself. An enemy arsenal under algorithmic control is only ‘theirs’ by contingencies of software dominance. From the military perspective – among others oriented to negative capability – the potential destructiveness of the technology is without determinable limit. Anything under software control falls into its realm. Which is to say that, asymptotically, everything does. But it doesn’t end there. AI also promotes an advance into virtuality.

Nuclear weaponry cuts a convergent path into purity of conception. No hydrogen bomb has yet been used against an enemy (or “in anger” as the singularly inappropriate expression goes). Thermonuclear warheads remain among a select category of virtual weapons, alongside a variety of chemical and biological agents, whose usage has been exclusively diplomatic, or even philosophical. The value of this military machinery is strictly counter-factual. Those ‘possible worlds’ in which they have been operationalized support little, if any, value of any kind. Weaponry supporting their potentiality floats the ontological option of extreme negative utility. They are – in the most rigorous sense – nightmare generators.

There is no reason (at all), then, to think that nuclear weapons are the last word in mass destruction. Nor can it be assumed that mass destruction is the ultimate criterion for deterrent weaponry. It is not only that high-energy physics opens a vast, rambling bestiary of virtual catastrophes which we have scarcely begun to peruse (although this is true). Physics has no monopoly on disaster, regardless of what its recent privileges might suggest.

It can never be a virtue for a weapon to be indiscriminate, which is to say imprecise. Turned around, we can say without hesitation or reservation that it is meritorious in any weapon, however absolutely devastating, for the greatest possible proportion of the damage it produces to be inflicted upon the enemy. In other words, a good weapon discriminates specifically against enemy interests. It hunts. There can be no serious doubt that the genomic biosciences and software engineering have more to contribute to this pursuit than physics possibly could.

Stuart Russell describes autonomous weapons as a “new, scalable class of WMDs.” The systems he is considering would be exemplified by drone swarms, “hunting in packs like wolves” (as one DARPA employee was indiscreet enough to reveal). Given enormous industrial production runs, performance specifications unshackled from human limitation, and targeting algorithms set for indiscriminate lethality, the devastating potential of such weapons would be hard to exaggerate. Their key, confidently predicted vulnerabilities, however, are at least as significant.

As Russell emphasizes, autonomous weapons could be subverted by a hostile “software update.” They could be hacked. Behind the menace of the hacker lies that of advanced artificial intelligence, mustering superior powers of cryptographic lock-picking and soft intrusion. Autonomous weaponry is therefore nested into a more profound threat.

AI designates a culmination of sorts. Nowhere else does destructive capability and rigid commitment promise to intersect more dynamically. Nothing separates the weapon from the game. It also counts, potentially, as an escalation.

Much criticism of the Cold War nuclear arms race already configured it as an existential risk, before the term had been coined. Between an X-risk and an extreme deterrent there no definite boundary. The difference is technical. Deterrence is a mode of employment. It uses negative utility. In this respect anything bad could be useful, were it not that a deterrent requires a trigger, under the control of the negotiating agent (at the point of negotiation). To threaten a potential aggressor with an asteroid strike makes no sense, unless an asteroid strike can be delivered. The same holds for geological disasters in general. All of which means that the acquisition of engineering capabilities on the largest scales, such as geo-engineering, weather control, climate regulation, and asteroid defenses – perhaps developed explicitly to avert potential existential risks – will inevitably expand the domain of deterrence options. In other words, techno-economic progress and the escalation of deterrence infrastructure are only formally differentiated. There is no materially persuasive way to improve the world that does not – on its occult side – widen the horizons of geopolitical horror.

Beside what could be had, there is the question of who has it. Beside the qualities of WMD-armed antagonists, their mere number is a source of terror, itself. It is only natural that multilateral deterrence should be found more threatening than its bilateral ideal, and now distant predecessor. Complexity scales nonlinearly in networks, and quickly becomes mathematically intractable. No one has any idea how massively distributed networks of insecurity would work. It is quite probably impossible to know. Deterrence is about to change phase.

Toothpaste doesn’t return to the tube just because it makes a mess. Once it is out, inconvenience has ceased to be any kind of argument against it. The dangers of a world in which ubiquitous deterrence capacity reigns are both obvious and immense. This is nevertheless the world we are entering. The trends driving it, from both the geopolitical and the techno-economic sides, are by any realistic estimation irresistible. Cheaper and more diverse nightmare weaponry is becoming available within an increasingly disintegrated international order. A variety of self-reinforcing dynamics – including but not restricted to those of the arms-race type – are further stimulating the process. Cascading acceleration is all but inevitable.

When conceived with maximal cynicism (i.e. realism), geostrategic independence is a direct function of deterrence capability. Don’t tread on me is the colloquial statement, whose perfect applicability is commonly under-estimated. The rattlesnake, combining fearsome weaponry with signaling, makes for a natural totem of deterrence. Neither venom, nor rattle, is dispensable. “Diplomacy without arms is like music without instruments,” runs the famous analogy, attributed to Frederick the Great. Game theory recognizes military capability as a communication medium.

It is not only that robust independence depends upon deterrence. Reciprocally, geostrategic liberty necessarily tends to the production of deterrence capability. An alien freedom, which could do anything, is – ineliminably – a threat. It provides the comprehensive model of the military threat. Whether ‘they hate us for our freedom’ or not, they have no choice but to fear us for it, and inversely. Geopolitics has no other origin. Any state without the will to scare also lacks the will to exist.

It’s all far more basic than we’ve been led to believe. As Niall Ferguson writes (realistically):

In the final analysis, borders are a function of power. If you can’t defend them, they are just dotted lines. The Kim dynasty’s calculation has been that nukes are the ultimate border guards. We shall soon find out if that calculation was correct. If so, many more states will want them.

Every geopolitical entity that is serious about sovereignty will want them, or something of at least equivalent deterrent credibility. The only alternative is naked dependency, made ever more uncomfortable by increasing global multipolarity, among the stark wreckage of any ‘world order’ or ‘international community’ grounded in the collective fantasy of miraculously authorized super-national norms. Explosive proliferation will be something the world has not seen before, even if it has already actually been there to see. We can be confident that the geopolitical order will be reconfigured by it.

What does explosive proliferation mean? Potentially, many things. For instance, vectors of technological – and thus economic – development are certain to be, to some significant degree, oriented by it. As artificial intelligence is factored into policy decision-making not only as an essential contributor to command, control, communications and intelligence (C3I), but as an intrinsic weapon of mass destruction, its prominence will be still further elevated.

WMD proliferation implies a multiplication of real geopolitical agencies. It is rigorously indistinguishable – in both directions – from a disintegrated world. Established relations of dependency are broken, releasing unanticipated – and evidently hazardous – freedoms. Whether or not this is the world we want, it looks increasingly inevitable that it is the world we are to have.