How nuclear war could start

Washington Post

a day ago

To understand how it could all go wrong, look at how it almost did.
If a nuclear war happens, it could very well start by accident.
A decision to use the most destructive weapons ever created could grow out of human error or a misunderstanding just as easily as a deliberate decision on the part of an aggrieved nation. A faulty computer system could wrongly report incoming missiles, causing a country to retaliate against its suspected attacker. Suspicious activity around nuclear weapons bases could spin a conventional conflict into a nuclear one. Military officers who routinely handle nuclear weapons could mistakenly load them on the wrong vehicle. Any of these scenarios could cause events to spiral out of control.
Such occurrences are not just possible plots for action movies. All of them actually happened and can happen again. Humans are imperfect, so nuclear near-misses and accidents are a fact of life for as long as these weapons exist.
The next nuclear age
This is the second article in a series by experts from the Federation of American Scientists examining why today's global nuclear landscape is far more complicated and, in many ways, more precarious than during the Cold War. Read part one.
Previous
Next
Today, nine countries have nuclear weapons. Most deploy them in peacetime on some combination of planes, missiles and submarines. Being ready and able to use them is seen as necessary to make them effective; more than 2,000 weapons are on alert and ready for use on short notice. So mishaps and accidents will continue to happen, with unpredictable results.
The extreme tensions of the Cold War fortunately never led to nuclear war, but the cocked-pistols deterrence of those decades produced plenty of close calls. Human error was a constant feature then and will remain an endemic risk in the new nuclear age — with more actors, faster and more complex technology, and more points of conflict and tension.
To understand how it could all go wrong, it is useful to look at some examples of how it almost did.
Blind man's bluff
HMS Vanguard sits in dock at a submarine base in 2006 in Helensburgh, Scotland. (Jeff)
In February 2009, a French ballistic-missile submarine (SSBN), Le Triomphant, carrying nuclear-armed strategic missiles, was conducting routine patrols in the Atlantic Ocean. At the same time, unknown to the French sub, the British HMS Vanguard SSBN, armed with its own nuclear-tipped missiles, was conducting a routine patrol nearby — too nearby. With 40 million square miles of open ocean, the two submarines collided.
Had either or both been sunk, there might have been no simple way to determine the cause. And in a time of crisis or tension — say, with Russia or China — the confusion could easily have escalated to crisis.
Advanced technology allows submarines to operate quietly, almost undetectably, in the vast space of the open ocean. This is accomplished through special propulsion systems engineered to be super-quiet and acoustic tile coatings on the hull that can absorb sonar waves, reducing the reflection that could make the submarine detectable. 'They make less noise than a shrimp,' the French defense minister said in explaining the collision.
Story continues below advertisement
Advertisement
Because of this, submarines are the least vulnerable of nuclear weapons, giving the nations that deploy them confidence they will always be ready. But what if one is suddenly lost? Military leaders know their subs would be logical targets for attack submarines as a precursor to a broader nuclear assault. If a nation believed its subs had been attacked, it might feel bound to retaliate.
Fortunately, in the 2009 collision, both subs were traveling at very low speed, no crew members were injured, and both the nuclear reactors and weapons remained undamaged in the incident. The French navy first reported its vessel had probably collided with a container ship. Though neither country was eager to discuss the incident, both submarines were visibly damaged. Still, they were able to return safely to their home ports. A member of Scotland's Parliament, Angus Robertson, reflected a general frustration when he said: The Defense Ministry 'needs to explain how it is possible for a submarine carrying weapons of mass destruction to collide with another submarine carrying weapons of mass destruction in the middle of the world's second-largest ocean.'
You might think this was a once-in-40-million accident, but it is by no means the only submarine collision between nuclear powers.
The Russian B-276 Kostroma. (Russian Navy) The nuclear-powered attack submarine USS Grayling on July 25, 1993. (OS2 John Bouvia/U.S. National Records and Archives Administration)
In March 1993, the USS Grayling, an attack sub operating near Russia's Kola Peninsula, struck the Novomoskovsk, a Russian Delta IV-class ballistic-missile submarine. The vessel, which could carry 16 ballistic missiles with an estimated 64 nuclear warheads, suffered a large dent in its outer hull. A similar incident occurred the year before when a U.S. Los Angeles-class submarine collided with a Russian Sierra-class attack submarine about 14 miles out of Russia's naval port in Murmansk.
That's because U.S. attack submarines routinely — and to this day — work day and night to find and track Russian strategic submarines, just as Russian submarines do the same to American, British and French vessels. Had this happened during a crisis or early phase of a conventional war, the accidental sinking of a ballistic missile submarine could well trigger an overreaction — and potential nuclear confrontation. (In these two cases, both sides immediately realized what had happened, and a crisis was averted.) Just imagine if such a collision had happened in October 2022, amid rumors that Russia was considering nuclear strikes in Ukraine.
Too hot to handle
Such risks are not just for nuclear superpowers. As recent events demonstrate, South Asia remains a potential nuclear flash point. U.S. officials initially sought to avoid any involvement in the clash between India and Pakistan in May, but the United States was eventually drawn into mediation when Indian drones struck Pakistan's Nur Khan air base in Rawalpindi, near the capital Islamabad. That base is also close to the headquarters of Pakistan's Strategic Plans Division, a key hub in the country's nuclear command and control system. U.S. officials apparently saw the strike and other events as risking escalation to the nuclear level.
Pakistan considers its nuclear weapons to be a last resort against a much larger Indian military, and it takes any threat to them very seriously. The escalating military action and rhetoric by two nuclear-armed rivals risked turning a conventional conflict into a nuclear crisis. The clash also saw large volumes of third-party disinformation, further clouding an already opaque and dangerous situation. And this is not the first such conflict between the two nations.
At 6:43 p.m. on March 9, 2022, Pakistan detected an Indian BrahMos cruise missile flying along the Indian side of the India-Pakistan border. After a few minutes of flight, the missile suddenly crossed into Pakistani territory and crashed about three minutes later into a hotel parking lot near the city of Mian Channu.
Without warning, a nuclear-armed state had just launched a missile into the territory of another nuclear-armed state. No one had any idea why.
Immediately, Pakistan reportedly placed military bases along the border on high alert, yet India kept silent, issuing no public statement and making no effort to communicate with Pakistan's military leaders.
Meanwhile, images and videos from the crash site began to pop up on social media. Pakistan held a news conference, releasing details about the event, asserting it was 'for the Indians to explain' what happened and why. Still, questions remained: How did this missile get launched? And was it an accident, or intentional?
Story continues below advertisement
Advertisement
Two days after the crash, India issued a public statement blaming 'a technical malfunction [that] led to the accidental firing of a missile.' The statement was revised two weeks later, this time blaming human error. Then, finally, in March 2024 — two years later — an Indian court revealed what really happened. During the visit of a high-ranking Indian Air Force officer, the launch crew — tasked with demonstrating a missile launch procedure to their boss — failed to disable the missile's combat connectors, leaving the weapon system ready for launch.
The error occurred during a period of relative peace between the two countries; no one was injured, and the site of impact was vacant. The Pakistanis quickly determined the missile was unarmed. No aircraft were hit, although several planes passed very close to the missile's flight path.
Errant missiles are more common than you might think. In November 2022, a missile killed two people at a grain facility in southern Poland, very near the border with Ukraine. For several hours, it was thought a Russian offensive missile had deliberately targeted a NATO ally, but in reality it was a Ukrainian air defense missile. For hours, Washington and NATO leaders feared the worst. When things get tense, even minor incidents can drive leaders to respond before all of the facts are known.
Red alert
The launch control center at the Minuteman Missile National Historic Site. (National Parks Service)
The end of humanity could arrive in minutes — that is what makes nuclear war so different from other wars. During the Cold War, both sides kept weapons on launch-ready alert, or 'hair trigger,' so as not to be caught by surprise. Fears of a 'bolt out of the blue' attack were real and enduring, and leaders thought the only way to deter such a strike was to be prepared to strike back in minutes. This inevitably led to errors and false alarms. Despite the best efforts of highly trained and capable military operators, mistakes are a fact of life in the nuclear world just as they are in regular life.
Between 1960, when the United States first deployed an early-warning system, and December 1976 — over 16 years — the system produced seven false alarms of nuclear attack. That's almost one every two years. Then, in 1979 and 1980, came five incidents in a row.
At 8:50 a.m. on Nov. 9, 1979, duty officers at NORAD and elsewhere were suddenly confronted with a realistic electronic display of a Soviet nuclear attack. The display indicated an offensive designed to decapitate the U.S. command system and destroy U.S. nuclear forces. A large number of Soviet missiles appeared to have been launched from land and sea.
Command post for North American Aerospace Defense Command operations, circa 1982. (U.S. National Archives and Records Administration)
This alert quickly passed Stage 1 — a missile display conference — when various anomalies are sorted out. It went to the next stage, a threat assessment conference. Strategic Air Command B-52 bomber crews were instructed to stand by, the Minuteman missile force was placed on low-level alert and the 'doomsday plane' was immediately airborne, providing the president with a secure way to order a nuclear strike even during an incoming attack.
Six minutes later, the threat assessment was called off — there were no signs of missiles in flight. It turns out that someone had mistakenly inserted a realistic exercise tape into the live warning system at NORAD headquarters.
Then, on March 14, 1980, early-warning systems picked up what appeared to be Soviet missiles headed for Alaska, Canada and the tip of Oregon. Again, the doomsday plane was rolling within minutes. It was another false alarm.
On May 28, 1980, early-warning systems suddenly displayed a Soviet missile barrage — one system showed 2,020 missiles launched. It was big, but brief, lasting only six seconds — so nothing was done in response.
On June 3 that year, the early-warning system reported two Soviet submarine-based missiles were launched, and the count quickly jumped to 200 missiles. Six minutes later, the systems reported 2,020 ICBMs — the same as that May 28 error. Twelve minutes later, the systems showed 200 submarine-launched ballistic missiles.
This all happened at 2:26 a.m. Bomber crews were told to start their engines, and a call was placed to President Jimmy Carter's national security adviser, Zbigniew Brzezinski, who later recalled that his military aide 'told me, flat out, we were under nuclear attack.'
The next step would have been to call the president and recommend retaliation.
After a harrowing 32 minutes, it was determined there were no missiles in flight. The alert was ended. Brzezinski did not awaken Carter, but the adviser was shaken. 'I had this feeling that we were all going to be dead in 28 minutes,' he later recalled. 'Basically, it was an all-out attack, and my recommendation to him ... would be — so, we fire on the warning.' The United States would retaliate 'with everything,' Brzezinski said. 'And my concern was, you know, how was I going to convince the president? How do I convince the president in the middle of the night when I am calling him that we are under attack, that I have already activated the Strategic Air Command, and I am urging him now to move for Option X? Suppose he says to me, 'Zbig, are you having nightmares? Zbig, are you crazy?' How do you do that with a president — an alert one, with naval experience?'
Three days later, on June 6, it happened again. This time 2,000 incoming missiles appeared on the warning display. Again, bomber crews scrambled, and engines were started.
The alert ended after 17 minutes when it was determined no missiles were in flight.
The May and June 1980 false alarms were caused by the failure of a single computer chip. A computer's peacetime message was supposed to continuously broadcast the digits zero-zero-zero, indicating there were no attacking missiles. The chip began inserting random 2s into the message, so it came out showing that 200 or 2,000 missiles were in flight.
That was not the only time computer issues caused unexpected alerts. Thirty years later, 50 Minuteman III nuclear intercontinental ballistic missiles suddenly went offline at Warren Air Force Base in Wyoming. Alarms designed to prevent theft or infiltration of the missiles also lost power. The failure was so serious that the president was briefed within hours.
The event turned out not to be a cyberattack from Russia, China or North Korea, but the result of an improperly installed circuit card during maintenance. But at the time, this was unknown and in a time of crisis, the event could have looked like a deliberate act of sabotage.
Story continues below advertisement
Advertisement
In 1983, the Soviet Union shot down a civilian Korean Airlines flight that had strayed over Siberia. A few weeks later, Soviet early-warning radars showed that a single U.S. ICBM had been launched toward the U.S.S.R. At a time of high tension, and given the fear of a U.S. first strike inside the Soviet leadership, such a launch could easily have triggered a massive counterattack. However, the watch officer, Col. Stanislav Petrov, had been trained that any U.S. attack would probably involve massive strikes, and he later stated that he considered a smaller strike — like the one his early-warning systems showed — to be illogical and therefore likely to be an error of some kind. He proved to be right. Would all Soviet watch officers have been willing to make the same call?
Another infamous event occurred in 2007, when U.S. airmen mistakenly loaded six live nuclear-armed cruise missiles onto a B-52 bomber instead of the inert training systems that were supposed to be used. The bomber flew across the United States to Barksdale Air Force Base in Louisiana, where it stood unguarded for a day until an officer discovered the missiles carried nuclear warheads. The Air Force was unaware it had loaded, flown with and lost control of six nuclear warheads.
In January 2018, President Donald Trump tweeted that his nuclear launch 'button' was much bigger than that of North Korean Supreme Leader Kim Jong Un. Days later, smartphones across Hawaii lit up with an emergency alert: 'Ballistic missile threat inbound to Hawaii. Seek immediate shelter. This is not a drill.' For almost 40 minutes, residents panicked. But an attack never came, because the message was a false alarm. 'An employee pushed the wrong button,' Hawaii Gov. David Ige (D) explained. Accidents happen all the time. Nuclear weapons are not immune from this dangerous reality.
What you can't see will hurt you
World leaders at the Group of 20 Summit in Osaka on June 28, 2019, prepare to take a joint photo. (Brendan Smialowski/AFP/Getty Images)
During the Cold War, the superpowers took pains to make sure their nuclear and conventional military capabilities were operated separately. Increasingly, that line is being blurred. The United States continues to keep its nuclear and conventional weapons separated, but it combines the command, control and communication systems for managing both and deploys aircraft that can carry either conventional or nuclear payloads. And the latest presidential nuclear guidance requires the military to integrate conventional and nuclear capabilities in its strike plans. In addition, Russia and China continue to develop dual-capable missiles that are designed to deliver both nuclear and conventional payloads that could be swapped without warning.
As a result, a nation being attacked might have no way to tell whether an incoming missile is nuclear or conventional. The difference in the response could be enormous.
Nowhere is this new danger as clear as in space. In November 2019, Russia launched a Soyuz rocket carrying a new type of satellite into orbit. Although Russia claimed that the spacecraft was intended to assess the 'technical condition of domestic satellites,' it soon became clear that its true purpose was different.
After remaining in orbit for less than two weeks, the satellite suddenly split into two. 'Like Russian nesting dolls,' noted Gen. John 'Jay' Raymond, then-commander of the U.S. Space Force. The two Russian satellites then began to tail a KH-11 U.S. spy satellite — one of four National Reconnaissance Office satellites providing coverage of the earth for the U.S. military. In an interview with Time magazine in February 2020, Raymond characterized this behavior as 'unusual and disturbing,' and having the 'potential to create a dangerous situation in space.' Russia had tested this 'nesting doll' technology three years earlier, during an instance in which one of the satellites also fired a projectile into space. Combined, these tests demonstrated the capability for Russia to track, trail and potentially hold U.S. satellites at risk.
This was the first time the United States military publicly revealed a direct adversarial threat to its satellites, and it wouldn't be the last. The unspoken but very real danger is that a satellite could be attacked — poking out the military eyes and ears of a country.
The United States relies heavily on satellites for nuclear command, control and communication, and they are the critical node allowing nuclear decision-makers to order the launch of their weapons. Yet many of those same satellites also have intelligence, surveillance and reconnaissance capabilities. They can detect and warn of incoming conventional enemy missiles. For those reasons, during a nonnuclear conflict, there could be strong incentives to disable an adversary's satellites. It would undermine that adversary's ability to see the entire battlefield and fully communicate with its weapons. And the adversary could conclude that such a hit was a prelude to a nuclear strike — and respond in kind.
As civilian and military assets in space are increasingly dual-use, it becomes more difficult for countries to discern intentions. For example, 'rendezvous and proximity operations' — maneuvers that bring spacecraft into close proximity with each other — have both civilian and military applications: They can be used to service and maintain satellites, and they can also be employed to disrupt another country's satellite operations. Ambiguities such as this will characterize the new nuclear age. Blurring of the lines between civilian and military assets is already complicated enough, but when it comes to weapons that can play both nuclear and nonnuclear roles, things can get especially worrisome. In a conflict, the co-mingling of nuclear and conventional forces could lead to a situation in which an attack against a country's conventional forces simultaneously threatens its nuclear arsenal.
Story continues below advertisement
Advertisement
Back on Earth, the mingling of nuclear and conventional weapons comes with real risks. Recently, Russia's use of what it called an experimental missile, the Oreshnik, in Ukraine showed yet another way this ambiguity can be dangerous. The missile carried a bundle of conventional warheads, but it can also carry nuclear weapons.
China's DF-26 intermediate-range ballistic missile has a 'hot swappable' warhead with a clamshell covering that can open so that a conventional warhead can be swapped for a nuclear warhead directly on the battlefield. This poses a serious challenge for U.S. military calculations. If Russian or Chinese dual-capable missiles were detected in flight, would the United States know whether the payloads were nuclear or conventional? And would the United States be able to target those dual-capable missiles without China or Russia assuming that a U.S. nuclear attack was incoming?
While a dual-use satellite going offline or a dual-capable missile being launched on a normal Tuesday probably wouldn't spark nuclear war, those same things happening in the midst of a nuclear crisis or an ongoing conventional war could rapidly bring the world to the brink.
Preventing catastrophe
There is no substitute for being able to reliably communicate with your adversary in a crisis to resolve a potential accident from spiraling out of control. Not only do all nuclear-possessing states need reliable ways of communicating with each other in a crisis, but these systems have to operate flawlessly when under attack. Ideally, they would also be available to local commanders.
Another critical element to avoiding catastrophic error or misunderstandings is transparency — through arms control agreements that provide a window into the forces and intentions of each country. These agreements enable predictability and reduce the risk of escalation. Legally binding and enforceable agreements are best, but even informal political agreements without verification measures have been useful in reducing risks. And verified arms control agreements can offer what spies and satellites cannot.
Story continues below advertisement
Advertisement
Leaders should not wait for a crisis to engage and manage these issues. During the cold war, the U.S. and Soviet Union faced an almost nonstop series of events that could have led to larger conflict. Often out of sight, mid-level military officers from East and West were in regular contact with each other on a host of nuclear and security issues. Today, connections are shrinking with Russia and are unreliably poor with China.
The United States and the Soviet Union had 30 minutes or less to make nuclear decisions. Those timelines will seem luxurious when missiles and underwater nuclear torpedoes can hit their targets almost without warning. Today's nuclear dangers demand that all states with these weapons take the steps necessary to prevent unwanted or accidental escalation — while there is still time.