Editor’s Note: Please also take a look at Tom Nichols recent piece Welcome to Russian Nuclear Weapons 101.

Nuclear war, the exchange of nuclear weapons between two or more states in open conflict. It’s unthinkable. It can’t happen.

Right?

Wrong.

Of course, nuclear war is extremely unlikely. Although the Bulletin of the Atomic Scientists has placed the hands of its famous clock at five minutes to midnight, that doesn’t mean very much and never has. The fact of the matter is that world nuclear inventories, led by reductions in the United States and Russia, have never been lower, and none of the major powers expects a nuclear conflict in the way they did during the Cold War. To crib a line from Captain Jack Sparrow, however, nuclear war is not impossible, it’s improbable, and a nuclear war could take place in more ways than you might think, sparked by any number of occurrences from a pure accident to an intentional strike.

I’m going to focus here on a war that could involve the United States and its allies on one side, and Russia or China on the other. Nuclear conflict between India and Pakistan, or between a future nuclear-armed Iran and Israel, is unlikely but far easier to imagine than a global nuclear conflict. Indeed, this is one reason Americans don’t think about nuclear war very much anymore: they think it will happen somewhere else. (If a regional limited war takes place, however, you’ll know it: even a small exchange of nuclear weapons will create a global environmental catastrophe that will dwarf Chernobyl or Fukushima.)

A small regional war, awful as it would be, would not destroy the United States nor threaten the end of the human race. A nuclear conflict of any serious size in the Northern Hemisphere, however, would effectively mean the end of the modern era. Further human progress would be subordinated to the basic needs of survival for years, if not decades, to come. A war between India and Pakistan would kill millions and pollute the earth for an eternity. But it would not threaten to bring the entire global system to a halt, or potentially lead to the release of thousands of warheads against of hundreds of cities across the globe, the “unthinkable” war for which Americans spent decades preparing, and for which we still maintain an arsenal of strategic weapons deliverable by air, land, and sea.

So how do we begin each of the nightmare scenarios?

1. Mechanical Accident:

The classic war-by-accident novel is Fail-Safe, in which a computer blows a fuse and the Air Force melts Moscow. Most people don’t read Fail-Safe anymore (which is why I make my students read it now), but almost everyone has seen WarGames and The Terminator, both of which are part of an entire genre of science fiction in which military computers decide not to take orders from pesky humans about who gets to fire nuclear missiles. That’s what makes them fiction, of course. Nuclear weapons are not sentient beings; Skynet is not self-aware and never will be. They’re just machines, and machines can fail on their own, with little human involvement.

As terrifying as it is to think of a war generated by a random mechanical hiccup, it’s important to note that this is the least likely trigger for a nuclear war. If anything, during the Cold War the superpowers spent so much time assuring the security of their arsenals against accidental use that both the Americans and the Soviets started to wonder whether they had too many barriers in place that could prevent the intentional launch of the weapons in wartime. While the danger of an accidental launch of a strategic nuclear weapon is not zero, it is tiny.

That is, unless someone builds a “Doomsday Machine” that takes the human beings out of the loop. And who’d be crazy enough to do that?

Turns out the Soviet high command, in its pathetic and paranoid last years, was just that crazy. The USSR built a system called Perimetr, known informally in Russia as “the Dead Hand.”  Perimetr was essentially a computer system that would watch for signs of nuclear attack and retaliate on its own if the Soviet leadership was struck first and wiped out. (I explained this is more detail for National Geographic, which you can watch here.)  We’ve since asked the Russians if it’s still on, and they’ve reassured us, with complete confidence, that we should mind our own business. Let’s hope they’re just being rude.

2. Human Error:

As long as there are machines run by human beings, there are going to be accidents. War, however, will not begin because a bomber crashes or a silo catches fire; rather, the error will lay in the misinterpretation of an accident by fallible human beings.

History is replete with such incidents. In 1995, the Russians forgot that the Norwegians had notified them of a rocket launch to put a weather satellite into space. The Russian high command told President Boris Yeltsin that they had a confirmed rocket launch from NATO over Russia. Fortunately, no one in the Kremlin assumed that Bill Clinton was trying to start World War III with a single warhead from Norway. Moreover, the warm relationship between Clinton and Yeltsin made the Russian president skeptical that Russia was under what Cold War strategists used to call a “BOOB,” or “Bolt Out Of the Blue” attack.

Similar mistakes have been provoked by flocks of birds, random computer glitches, and the sun glinting off cloud formations (which was interpreted by Soviet computers as the fiery tails of multiple U.S. missile boosters). In each case, it was up to a human being to make the call: is someone really attacking us? Smart people in both Russia and the United States have prevented these mechanical errors from turning into Armageddon.

Nevertheless, the declassified files on these incidents won’t exactly help you sleep more soundly. In 1979, for example, NORAD, the joint U.S.- Canadian North American Air Defense Command, rousted White House advisor Zbigniew Brzezinski out of bed and told him that a massive Soviet nuclear strike was incoming. Or so they thought, anyway: they were giving him a heads-up while they checked it out. Brzezinski was minutes away from waking President Jimmy Carter and handing him the codes to Hell when NORAD called back and said: Oops, nevermind. The computers goofed. Our bad. We’ll fix it.

And then it happened again in 1980.

The Soviets, who got wind of all this activity, politely sent a note to Carter asking him, in effect: What the is going on over there? It was a good question, and we’d have asked the same thing.

The key here, however, is not the technology, which makes a Fail-Safe type accidental launch nearly (but not entirely) impossible.  Rather, the danger lies with human beings, who could issue orders to retaliate in moments of duress that cannot be undone. While this is less of a risk when tensions are lower, it’s always a possibility. Combining mechanical error with the natural flaws of human judgment multiplies the risk of accidental war from an infinitesimal risk to a very real possibility.

3. A Show of Force:

As we move from mechanical errors to human agency, things actually get scarier. Machines can make mistakes, but absent an international crisis and additional confirming evidence, no one goes to war on the say-so of a malfunctioning HAL 9000. While journalists and nuclear safety experts have written some excellent books about accidental detonations and other risks, I worry far more about a conscious decision to use nuclear weapons.

The worst mistake to make about nuclear weapons is to believe that they are ordinary arms, available for military use like any other. (This is sometimes called the “conventionalizing” of nuclear weapons.) The second worst mistake, however, is to believe that nuclear weapons are magic, and that using them solves problems that are otherwise politically or strategically intractable. This second error is what leads people into thinking about things like “demonstration shots” or nuclear shows of force, in which a nuclear weapon is exploded near, but not in, a conflict.      

Ostensibly, such a dramatic demonstration will bring all of the combatants to their senses and call off whatever started the whole ruckus in the first place. (Nuclear game-players, by the way, almost never tell us why we’re at the brink. Politics is a messy business that just complicates clean, beautiful models and equations.) This idea is both seductive and dangerous: if our enemy sees our resolve, the logic goes, he will cease his predations. It is possible, of course, that a nuclear explosion could very well focus the attention of Russian, Chinese, or American leaders and invoke a “moment of clarity,” where everyone thinks very hard about what’s at stake and how much they’re willing to risk over it.

But to count on such clarity is a huge and potentially disastrous gamble. What happens if a demonstration of resolve merely provokes a corresponding act from the other side? One flexing of nuclear muscle (at sea, perhaps, or far from population centers on land) then leads to another. No one backs down, and so the next shot is a little closer to something that matters. Or perhaps a human being – again, always the weakest link in the nuclear chain – interprets a “demonstration” as the prelude to full-blown attack, and decides that safe is better than sorry. When it comes to strategic nuclear exchanges, there is no prize for second-place. The temptation to act, barely restrained during a crisis, might become overwhelming.