From invading animals to a faulty computer chip worth less than a dollar, the alarmingly long list of close calls shows just how easily nuclear war could happen by mistake.

It was the middle of the night on 25 October 1962 and a truck was racing down a runway in Wisconsin. It had just moments to stop a flight.

Mere minutes earlier, a guard at Duluth Sector Direction Center had glimpsed a shadowy form attempting to climb the facility’s perimeter fence. He shot at it and raised the alert, fearing that this was part of a wider Soviet attack. Instantly, intruder alarms were ringing at every air base in the area.

The situation escalated remarkably quickly. At nearby Volk Field, an air base, someone flicked the wrong switch – so rather than the standard security warning, pilots heard an emergency siren telling them to scramble. Soon there was a frenzy of activity, as they rushed to take to the skies, armed with nuclear weapons. 

It was the height of the Cuban Missile Crisis and everyone was on edge. Eleven days earlier, a spy plane had captured photographs of secret launchers, missiles and trucks in Cuba, which suggested the Soviets were mobilising to strike targets across the United States. As the world knew only too well, all it would take was one single strike from either nation to trigger an unpredictable escalation. 

As it happens, on this occasion there was no imposter – at least, not a human one. The figure skulking around the fence is thought to have been a large black bear. It was all a mistake.

But back at Volk Field, the squadron was still unaware of this fact. They had been told there would be no practice runs, and as they boarded their planes, they were entirely convinced that this was it –World War Three had begun.

In the end, the base commander figured out what had happened. The pilots were intercepted by a quick-thinking official, who drove a truck at them as they started their engines on the runway.

Fast-forward to today and the atomic anxiety of the 1960s has all but been forgotten. Nuclear shelters are the preserve of eccentric survivalists and the ultra-rich, and existential worries have shifted to other threats such as climate change. It’s easy to forget that there are roughly 14,000 nuclear weapons out in the world, with the combined power to extinguish the lives of around three billion people – or even the extinction of the species if they triggered a nuclear winter. We know that the prospect of any leader intentionally detonating one is extremely remote; after all, they would have to be mad.

What we haven’t factored in is that it could happen by accident.

All told, there have been at least 22 alarmingly narrow misses since nuclear weapons were discovered. So far, we’ve been pushed to the brink of nuclear war by such innocuous events as a group of flying swans, the Moon, minor computer problems and unusual space weather. In 1958, a plane accidentally dropped a nuclear bomb in a family’s back garden; miraculously, no one was killed, though their free-range chickens were vaporised. Mishaps have occurred as recently as 2010, when the United States Air Force temporarily lost the ability to communicate with 50 nuclear missiles, meaning there would have been no way to detect and stop an automatic launch.

Despite the staggering cost and technological sophistication of modern nuclear weapons – the US is expected to spend $400bn (£306m) on its capabilities between 2017 and 2026 – the historical record tells its own story and shows just how easily the safeguards we set up can confounded by human error or curious wildlife. 

On 25 January 1995, the then-Russian President Boris Yeltsin became the first world leader in history to activate a “nuclear briefcase” – a satchel which contains the instructions and technology for detonating nuclear bombs.

Yeltsin’s radar operators had noticed that a rocket had been launched off the coast of Norway, and they watched uneasily as it rose into the sky. Where was it headed – and was it hostile? With the briefcase in his hands, Yeltsin frantically conferred with his top advisors about whether to launch a counter-attack. With minutes to decide, they realised it was headed out to sea and was therefore not a threat.

It later emerged that this was no nuclear strike, but a scientific probe, which had been sent to investigate the northern lights. Norwegian officials were left baffled that it caused such a commotion, because the launch had been publicly announced at least a month earlier.

Crucially, it doesn’t matter whether a nuclear strike is initiated because of a mix-up or a real threat – and once it’s been sent, it’s irreversible. “If the president responds to a false alarm, then he will have accidentally started a nuclear war,” says William Perry, who served as the US Secretary of Defense under former President Bill Clinton and the Undersecretary of Defense for the Carter administration. “There's nothing he can do about that. The missiles cannot be called back and they cannot be destroyed.”

Why have there been so many perilously close shaves? And what can we do to prevent another one happening in the future?

How nuclear attacks unfold

At the root of the potential for mistakes are the early warning systems set up during the Cold War. Instead of waiting for nuclear missiles to strike their target – which would, naturally, provide concrete proof of an attack – these aim to detect them early on, so that a retaliation can be launched before their own weapons are destroyed. 

To do this, you need data. 

Unbeknown to many Americans, the US currently has a number of satellitessilently watching over it at all times, including four which operate from 22,000 miles (35,400 km) above the Earth. They’re in “geosynchronous orbit” – located in a sweet spot where they never change position relative to the planet they’re circling. This means they have a more or less constant view of the same area, so they can detect the launch of any potential nuclear threat, seven days a week, 24 hours a day.

What satellites can’t do is track a missile as it travels. For this, the US also has hundreds of radar stations, which can determine their position and velocity, and calculate their trajectories.

Once there are enough indications that an attack is underway, the president is informed. “So perhaps about five to 10 minutes after the launch of the missiles, the president will get word of that,” says Perry. They have the unenviable task of deciding whether to strike back.

“It's quite a complicated system and it works nearly all the time,” says Perry. “But we’re talking about a low probability, high consequence event.” It only needs to happen once.

Wayward technology

There are two kinds of error which can lead to a false alarm – technological and human (or, if we’re really unlucky, both at once). A classic example of the first happened while Perry was working for US President Jimmy Carter back in 1980. 

“It was such a shock,” says Perry. It started with a phone call at 3:00, during which the watch office of the US air defence command informed him that surveillance computers had discovered 200 missiles headed directly from the Soviet Union to the United States. By then, they had already realised that it wasn’t a real attack – the computers had somehow got it wrong. 

“They actually had called the White House before they called me – they called the president. The call went through to his national security advisor,” says Perry. Luckily he delayed for a few minutes before waking the president, during which time they received the information that it was a false alarm.

But if they hadn’t waited – if they had woken Carter up immediately, the world could be a very different place today. “Had the president got the call himself, he would have had about five minutes to decide whether to launch or not. This is the middle of the night, no chance to consult anybody,” says Perry. From this moment on, he never thought of the prospect of a mistaken launch as a theoretical problem – this was a genuine and alarmingly realistic possibility. “I would say it was very close,” he says.

On that occasion, the problem turned out to be a faulty chip in the computer running the nation’s early warning systems. It was later replaced for under a dollar. A year earlier, Perry had experienced another close shave when a technician inadvertently loaded the computer with a training tape, and accidentally broadcast the details of a highly realistic (but nonetheless fictional) missile launch to the main warning centres.

Which brings us to the issue of involving the deeply flawed brains of bipedal apes in a process involving weapons with the power to flatten entire cities. And clumsy technicians aside, the main people we have to worry about here are those who actually have the power to authorise a nuclear strike – world leaders.

“The US president has complete authority to launch nuclear weapons and he's the only one that does – sole authority,” says Perry. This has been true since the days of President Harry Truman. During the Cold War, the decision was delegated to military commanders. But Truman believed that nuclear weapons are a political tool and therefore should be under the control of a politician.   

Like all those who went before him, President Donald Trump is followed everywhere he goes by an aide carrying the nuclear “football”, which contains the launch codes for the nation’s nuclear weapons. Whether he’s up a mountain, travelling in a helicopter or sailing across the ocean, Trump has the ability to launch a nuclear strike. All he has to do is say the words and mutually assured destruction – “MAD”, where both the attacker and the defender are totally annihilated – could be achieved within minutes.

As many organisations and experts have pointed out, concentrating this power within a single individual is a big risk. “It's happened a number of times that a president has been heavily drinking, or subject to medication he's taking. He may be suffering from a psychological disease. All of these things have happened in the past,” says Perry.

The more you think about it, the more disturbing possibilities emerge. If it’s night time, would the president be asleep? With minutes to decide what to do, they’d barely have time to regain consciousness, let alone refresh themselves with a cup of coffee; it seems unlikely they’d be functioning at their highest level.

In August 1974, when US President Richard Nixon was embroiled in the Watergate scandal and on the cusp of resigning from office, he became clinically depressed and emotionally unstable. He was rumoured to be exhausted, regularly binging on martinis and generally behaving strangely – a Secret Service agent apparently once saw him eat a dog biscuit. Nixon had reportedly always been subject to rage, drinking and potent prescription drugs, but this was much more serious. And yet, he still had the power to launch nuclear weapons.

(Intoxication is also a problem among the military personnel who guard the nation’s nuclear arsenal. In 2016, several US air-crew working at a missile base admitted to taking drugs including cocaine and LSD, and four were later convicted.)

How to avoid a catastrophic accident 

With all this in mind, Perry recently co-authored a book – The Button: The New Nuclear Arms Race and Presidential Power from Truman to Trump – with Tom Collina, director of policy at the nuclear non-proliferation charity Ploughshares Fund. In it, they outline the precariousness of our current nuclear safeguards, and suggest some possible solutions.

First, up, they would like to see an end to sole authority – so that decisions about whether to launch these weapons of mass destruction are made democratically, and the impact of any mental impairments on the decision is diluted. In the US, this would mean holding a vote in Congress.

“This would slow down the decision about whether to launch them,” says Perry. It’s commonly assumed that a nuclear response must happen quickly, before the ability to strike back is lost. But even if many cities and all land-based missiles in the US were wiped out by nuclear weapons, the surviving government could still authorise military submarines to launch. “The only kind of retaliation that’s warranted is one where you know they are attacking. We should never respond to an alarm that could be false,” says Collina. And the only truly reliable way to make sure a threat is real is to wait for it to land.

Moving at a calmer pace would allow countries to keep the deterrent benefits of mutually assured destruction, but with a significantly lower chance of blundering into a nuclear war because, say, a bear is climbing over a fence.

Secondly, Perry and Collina make the case for nuclear powers pledging to only use nuclear weapons in retaliation – and never being first. “China's an interesting example, because it already has a no first use policy,” says Collina. “They've announced that they will not use nuclear weapons first in a crisis, and there's some credibility in that policy because China separates its warheads [which contain the nuclear material] from its missiles [the delivery system].” This means China would have to mate the two together before launching an attack, and with so many satellites constantly watching for this, presumably someone would notice.

Notably, the US and Russia have no such policy – they reserve the right to launch their nuclear weapons, even in response to conventional warfare methods. Adopting “no first use” was considered by the Obama administration, though they were never able to reach a decision. 

Finally, they argue that it would be beneficial for countries to retire their land-based intercontinental ballistic missiles entirely, because they could be destroyed by an incoming nuclear attack, these are the weapons that are most likely to be deployed in haste if an attack is suspected, but not confirmed.

Another possibility would be to enable nuclear missiles to be cancelled, in the event that a provocation turns out to be a false alarm. “It's interesting because when we do test flights, they can do this,” says Collina. “If they go off course, they can self-destruct. But we don't do that with live missiles, because of the fear that an adversary will somehow get the remote controls and be able to disable them.”

And there are other ways a country’s own technologies could be used against them. As we become more and more reliant on sophisticated computers, there is growing concern that hackers, viruses or AI bots could start a nuclear war. “We believe that the chance of false alarms has gone up with the increased danger of cyber-attacks,” says Collina. For example, a control system could be spoofed into thinking that a missile is coming, which could mean a president is tricked into launching a counter-attack.

The wider problem, of course, is that nations want their nuclear weapons to be rapidly responsive and easy to use – available at the push of a button. This inevitably makes it harder to rein in their use.

Though the Cold War is long gone, Collina points out that we’re still set up for a “bolt out of the blue” unprovoked attack – when in reality, we now live in a radically different world. Ironically, many experts agree that by far the biggest threat comes from the very launch systems that are supposed to be protecting us.

Zaria is a science writer based in London, though her path to journalism has been far from typical.

The original article was published on BBC.

More about: nuclear