Simon Beard has recently published an article in BBC History Magazine on historical cases in which humanity narrowly avoided a global catastrophe, mainly due to nuclear war.

Working in the field of human extinction it can be hard to stay focused. Global Challenges, from the future of AI to the catastrophic effects of global warming, seem to be everywhere nowadays and we are often being asked to comment on this week’s big fear. Recently, of course, there has been considerable renewed focus on the risks of nuclear weapons, with many now asking, just how worried should we be about a nuclear war between the USA and North Korea.

For once, I was very glad of this media focus, because this is an issue I had been studying over the summer. In the 70 years that have elapsed since the first nuclear weapons were developed, history has actually provided us with a considerable evidence base about possible ways that nuclear wars might begin.

In truth many of these incidents are cases of human error rather than Hollywood brinkmanship: the Russian nuclear missile commander during the Cuban missile crisis who was not told of the USA’s plans to drop practice depth charges in the area he was patrolling, the engineers who accidently loaded an exercise tape into the USA’s NORAD missile detection system simulating a Soviet attack, the British and French Nuclear Submarines that collided under the Atlantic Ocean, and that sort of thing. Some of these errors occurred in part because of heightened tensions between nuclear powers, but many came out of a clear blue sky. Had any of them escalated into a full blown nuclear conflict, not only would the world not know what hit it, but people would not have had a clue why either.

In other cases there was at least some kind of tactical game afoot, most notably during the Cuban Missile Crisis in 1962 and the Cargill Crisis between India and Pakistan in 1999. However, there are other incidents that are less well known, such as Operation Able Archer, a 1983 live fire NATO military exercise that the USSR mistakenly believed to be real military preparations for a nuclear war, only for their own military response to be mistaken by NATO for just another military exercise.

The more I studied these incidents, the more I concluded that we respond to them in the wrong ways. There is a tendency to see each and every case as a terrible mistake, a one-off freak accident that must never be allowed to happen again. However, when you see so many different incidents lined up side by side you realise that this just true. Nuclear near misses are not unconnected moments of madness, they are a systemic feature of our ability to do such massive damage in such a short period of time.

For instance, colleagues at the Global Challenges Foundation have mapped out all the various ways in which a nuclear war could start, their conclusions look like this:

Most of the stages on this tree have happened at some point or other during the past 70 years, and many of them seem to occur frequently enough for us to conclude that there is a statistical pattern in their occurrence (especially those events that rely on human error). What saves us from the disaster of a nuclear war is that the probability of each stage cascading up the tree into something more serious is actually quite low. It seems that, whether people are being faced with a false alarm or an intentional provocation, cool heads tend to prevail and a crisis is avoided. This is good news.

What is less good news, is that with so many nuclear near misses occurring in recent history, even a low probability that the next incident will not escalate is too high for comfort.

So far there 4 different attempts have been made to quantify just how low this probability might be. In 2008 Martin Hellman estimated that the annual probability of a ‘Cuban Missile Type Crisis’ producing a nuclear war was 0.02% to 0.5%. Then, in 2013, Anthony Barrett and colleagues at the Global Catastrophic Risk Institute estimated the annual probability of a nuclear war between the USA and Russia as being somewhere between 0.001% and 7%. Carl Lundgren has estimated that over the past 66 years we have faced an annual probability of nuclear war that was greater than 1% per year. Finally, in 2015, Dennis and Armstrong surveyed expert opinion to estimate that there is an approximately 0.05% chance per year of a nuclear war that had the potential to cause human extinction. There is a lot of variation and uncertainty here, but given the different approaches and methodologies being used, we can say with some confidence that the probability of a nuclear war starting is likely around 0.1% per year, and that it could be considerably higher.

What does this mean? Well let’s assume that if there was an international nuclear war, your chances of being killed would be around 10%. If the annual probability of such a war is 0.1%, this gives you a 1 in 10,000 risk of dying in a nuclear war each year, or a 1 in 125 risk of being killed this way over the course of an 80-year lifespan. That is about as high as your risk of dying in a motoring accident (and incidentally means that, even in a country like the USA, people are more likely to be killed by a nuclear warhead than a firearm).

So, the next time you read about rising tensions between two nuclear armed states, remember that this may be an international car crash in more ways than one and you should probably try to worry about things escalating about as much as you would if you saw somebody driving dangerously. Sadly, such incidents aren’t one off tragedies caused by the unique personalities of those involved, they are a regular fact of life and deserve sustained effort to prevent and avoid. As long as states keep hold of their nuclear weapons, this is how things are likely to remain.



Tel: +44 1223 766886