As I described in my previous piece ^[1], I’ve recently come around to the view that we need to do something to reduce the probability of a civilisation-ending nuclear exchange. Put like that, it sounds like a no-brainer, but I received quite a bit of pushback from strategists I sent a draft to. I’ll respond to the arguments they raised here.

As foreshadowed in part 1, there was the predictable argument that the world has become safer since the advent of nuclear weapons, because they have appreciably reduced the incidence of major conflict. I would have said that once as well, but looking at the problem through a statistical lens, I don’t believe it’s true. While we haven’t seen violence on the scale of the two world wars, that’s a very poor measure of relative peace. As Lewis Richardson points out in his statistical analysis of wars, those are the only two ‘magnitude 7’ wars (i.e. more than 10 million deaths) in human history. Saying we’re safer now than we were in World War II is a bit like saying the earth is cooling because last year was slightly less hot than the previous year. The temperature reduction between 2016 and 2017 is true, but in isolation it doesn’t accurately reflect the underlying dynamics ^[2].

There’ll be excursions around the mean in any noisy dataset. Interpreting those as systemic trends is incorrect. Since World War II, there have been as many magnitude-6 wars (over a million deaths) as in previous half-centuries, including both interstate wars (Korea ^[3] and Vietnam ^[4]) and civil wars (China ^[5] and Rwanda ^[6]). There’s a magnitude-5 war underway at the moment in Syria, and the conflict in Yemen could easily reach that level. And many of the Cold War conflicts involved proxies of the two competing blocs—as mirrored in the Syrian conflict today. Time will tell if nuclear weapons have really calmed things down, but I’d be surprised if the apparent current period of relative peace is statistically significant—though the increased global population might have reduced the per capita risk of death in conflict.

Another popular argument was that the near misses of the Cold War provide confidence that we can safely negotiate crises. That’s deeply worrying. Ellsberg points out that if one or two individuals had acted differently at key moments, we wouldn’t be here having this argument today. Two especially close calls were on board the Soviet submarine B-59 during the Cuban missile crisis, when the decision to not fire a nuclear-tipped torpedo came down to one man ^[7] overriding another, and the ‘Able Archer’ NATO exercise of 1983, which almost set off a Soviet nuclear strike. (You can read Paul Dibb’s account of the 1983 event here ^[8].)

To take comfort from those events (and others, including the moon being mistaken for an incoming Soviet missile ^[9] with over 99% confidence), you have to argue that there’ll always be one or more of the right people in the right place at the right time to intervene to head off disaster. Instead, those incidents are evidence of the very substantial risks that exist, and of the very thin margins by which we have managed to not realise those risks so far. When playing Russian roulette, two successive ‘clicks’ shouldn’t provide greater confidence that future trigger pulls will also end happily. To get quantitative, if the odds of a crisis event being mismanaged is as low as 10% (and that seems pretty generous in some of the documented instances), then on average we get to roll the dice just six times before disaster strikes.

A related argument was that delegation of nuclear responsibility is actually a good thing when the leadership of nuclear-armed states is unreliable, such as mercurial leaders like Donald Trump and Kim Jong-un. In that case, one respondent argued, it’s good to have the possibility of someone else saying ‘no’. But that relies on level heads being in all of the right places further down the chain of command on both sides. And having inexpert leaders who are prone to creating crises surely outweighs the dubious benefits of a longer command chain.

I did get one argument that’s more convincing—that ‘you can’t get there from here’. The current reality is that nuclear weapons exist, and states in possession of them aren’t likely to agree to do away with them, especially when that creates a strong incentive for others to cheat on the deal. That’s likely true, but it doesn’t mean that we have to settle for a perpetual existential threat. Ellsberg convincingly explains that a single nuclear detonation in Washington or Moscow, regardless of its origin, could be the beginning of the end.

There are policy approaches that could substantially reduce the risk of global catastrophe. For a start, we should continue to take every possible step to prevent further proliferation of nuclear weapons. More fingers on more triggers would be worse. And that goes for us—Australia’s position should be ‘no nukes’, regardless of any possible future degradation of the US security guarantee. Another approach might be an international treaty that prohibits fusion weapons and limits yields to kilotons rather than megatons. Given the accuracy of modern delivery systems, large-yield weapons shouldn’t be required for nuclear deterrence to hold. (Reducing the total world holding to levels below that required to produce a nuclear winter is not a new idea, and dates back to the 1980s ^[10].)

That said, I’m not holding my breath. Humans don’t seem to be particularly good at dealing rationally with existential threats. I’ll give the last word to one of my strategic studies students from ANU. I was discussing the management of risk, and offered a large cometary impact as an example of a low likelihood but extreme outcome event (ask T. rex how that worked out). I speculated about the global response if astronomers spotted a comet with a 5% probability of hitting earth in the next century. The student’s comment was that, ‘If global warming is anything to go by, comet denial would become a thing.’ It’s darkly funny, but I think we’ve also been in nuclear denial for a long time.

 

Further reading

Daniel Ellsberg, 2017, The doomsday machine: confessions of a nuclear war planner ^[11].

Brian Hayes, 2002, ‘Computing science: statistics of deadly quarrels’ ^[12], American Scientist, vol. 90, no. 1 (summary of findings in Lewis F. Richardson, 1960, The statistics of deadly quarrels (out of print)).

Richard Rhodes, 2008, Arsenals of folly: the making of the nuclear arms race ^[13].