Accidental Nuclear War: Don’t Lose Sleep over It

May 22, 2023

Recently, I got into a fruitful exchange on Twitter about the dangers of accidental nuclear war during the Russo-Ukrainian conventional war. Since this is an argument / fear I hear expressed quite often, I thought it would be worthwhile to explain why the risks of this particular road to nuclear war are very low indeed. In fact, they are so low that I think that the risk of a nuclear missile not flying when ordered to is much higher than the risk of one flying when not ordered to.

An accidental nuclear war is one started by an unauthorized / inadvertent launch of nuclear weapons that could be the result of a mechanical malfunction, miscalculation, or loss of control. There are several pathways that are supposed to lead to such war, so let’s look at each briefly.

Peacetime accident. In January 1963, Secretary of Defense McNamara said that a full-scale nuclear war could be triggered by an accident similar to the crashes in Texas and North Caroline of American aircraft carrying nukes. In the NC crash, all but one of the safety systems on the weapon failed, and there could have been a nuclear explosion on US soil. There were at least 40 accidents involving nuclear weapons that we know of during the Cold War — some of them resulting in warheads being lost at sea or contaminating significant areas — with the West “leading” with 6 out of 10 worst air-related ones, and the Soviets “leading” with submarine problems.

It is not exactly clear how such an accident would escalate to a nuclear war though. Perhaps one could imagine that in the confusion surrounding the news of a nuclear explosion in America and the government going on full alert until the situation is clarified, an adversary might be frightened and prepare as well, but it’s a far-fetched scenario. It’s extraordinarily unlikely that the US government would order retaliation of any sort until making sure one is necessary. The reason is simple: there have been at least 40 accidents, everyone knows that they are possible despite precautions, and so nobody would go off on a hair-trigger, especially during peacetime.

In fact, even in the most intense crisis that we have had — the Cuban Missile Crisis of 1962 — both sides bent over backwards to interpret each other’s behavior in the most charitable terms possible precisely because both wanted to avoid escalation. When the crisis began, Strategic Air Command installed nuclear warheads on nine of the ten ICBMs at Vandenberg AFB in California. At the height of the crisis, they launched the tenth missile on a previously scheduled test over the Pacific — somebody had neglected to cancel it. The Soviets could have interpreted it as a prelude to attack, but they did not. Similarly, when the Soviets shot down the U-2 plane over Cuba, the US government did not know whether it had been authorized by Khrushchev or not, and decided to act as if it had not been, so they ignored it.

Recently, when a missile — thought to be Russian — landed in Poland and killed two farmers, a lot of people started clamoring for activating Article V of NATO, and some made dire predictions about dangers of escalation. All governments held firm until investigations were completed, and then concluded that it had been an accident, most likely a Ukrainian missile that had been trying to take out a Russian one. No retaliatory measures were taken and the restrained reaction of the governments is something one should expect.

False alarm. The American public used to be frightened by journalists penning articles about how a flock of birds confused the computers, which sent bombers scrambling and put missiles on alert. Early warning systems are tricky because they have to be sensitive enough to detect incoming missiles and yet not too sensitive to trigger false alarms. Of course, computers have advanced significantly since but even now, nobody is going to order a nuclear strike on the warning of one radar station. Electronic hallucinations of the early warning system would have to be confirmed by other, more reliable, systems. Computers these days monitor everything from aircraft movements, bird migration patterns, orbits of devices in space, and are much less likely to get confused by non-missile events. As with accidents, because people know that false alarms are possible, every effort is made to ascertain the veracity of reports before making final decisions.

The military would, of course, prepare for retaliation immediately, and this could include launching aircraft, bringing missiles into state of readiness for launch and so on. The nightmare scenario that people like to talk about is “launch on warning,” which is the policy of launching a retaliatory strike before a single enemy nuclear weapon had exploded. From a military standpoint, this policy makes sense as explosions in the atmosphere could result in an electro-magnetic pulse (EMP) with sufficient coverage to damage communications and electrical systems, partially incapacitating the US. Bruce Blair, the author of The Logic of Accidental Nuclear War, has been a vocal critic of the launch-on-warning policy. Incidentally, he was mostly concerned with the Russians accidentally launching something because of the deterioration of their early warning system. As an example, consider the events of January 25 1995, when Norwegian and American scientists launched a rocket near Spitsbergen to study the aurora borealis. The rocket crossed a flight corridor that could be used by ICBMs stationed in North Dakota to attack Moscow, and the Russian early warning system detected it. When the rocket reached the altitude used by submarine-launched Tridents, the Russians went on high alert fearing a nuclear attack, and President Yeltsin had to consider retaliation. The Russians continued to monitor events and concluded that no attack was taking place, and stood down their nuclear force.

Several false alarms were triggered during the Cuban Missile Crisis as well. On October 28, NORAD was notified that a nuclear missile had been launched from Cuba and was heading for Tampa, FL. They waited, in part because they thought that a single launch was unlikely, and it turned out that a radar operator had mistakenly inserted a test tape simulating a nuclear attack into the system, causing the control room officers to trigger the alarm. There was another false alarm when an inexperienced operator interpreted what he was seeing incorrectly and thought two missiles were headed for the US. Again, the military waited for confirmation before taking any action.

Another notorious incident took place on November 9 1979, when the US early warning system showed 1,400 Soviet ICBMs en route to North America, again because of a computer simulation mistakenly started at NORAD at Cheyenne Mountain. Another one on June 3 1980 showed 200 SLBMs and 2,020 ICBMs heading for the US. (Recent disclosures show at least four false alarm during 1979-80.) The 1979 incident alarmed the Soviets too, who lodged a formal complaint warning of the risks involved. While it is surely distressing that 17 years after the “original” computer simulation accident the US military was still using the same computers to run simulations and do real-life monitoring (testing has since been moved off-site), the main thing here to remember is that in all cases, the US leadership always did the same thing: it waited for confirmation.

It is not easy to imagine a realistic scenario where a nuclear strike would be ordered because of false alarms. In my exchange on Twitter, the best scenario described involved the coincidence no fewer than three unlikely events: (1) cyber attacks that affect the early warning system, causing it go haywire, (2) loss of communications between command that could be pre-authorized to launch and political/military leadership; (3) trigger-happy commanders who launch because they misinterpret the loss of communications as part of incoming nuclear strike. Setting aside the question of which hostile power would be capable of disrupting our early warning system with a cyber attack and why they would want to do this if it could trigger a nuclear response, let’s just say that all experience suggests that this scenario is so far-fetched that it would be rejected as a script for a Hollywood movie as being too unrealistic.

Unauthorized launch. This one is a real possibility, and we have had several “near-misses,” as did the Soviets and the Russians after them. The issue here is the potential loss of political control in one of three scenarios: (1) local commander goes crazy and launches without authorization; (2) local commander under pressure of crisis retaliates with nuclear weapons; (3) military leadership overreacts and sends things spiraling out of control despite efforts of civilian leadership to restrain it.

The system of checks, safeguards, and fail-safe mechanisms designed to prevent these is fairly extensive. There are tests and background checks to screen out unsuitable candidates (this includes ones that would not follow orders to launch as well), and reliability assessments are repeated at regular intervals to detect any changes in disposition. There are physical safety devices that make it exceedingly difficult for a single person or even a group to launch without authorization. The steps involve precise sequences that have to be done in time-limited intervals, and any failure requires a restart of the process from a high level. Because the physical takeover of a facility remains a possibility, nuclear warheads are normally stored separately from their delivery systems, and transported in secret and under heavy guard only when absolutely necessary. In peacetime, there’s virtually no chance of a launch anymore (the situation was different early on during the Cold War).

Things are different during a crisis, however, when the “defense readiness condition” (DEFCON) reaches a level that the nuclear arsenal is put on alert status and is ready for use (this is when the nuclear warheads would be installed on the missiles, for example, and the strategic bombers might be on continuous airborne alert). Although the DEFCON level is not public, it is believed that right now it is at DEFCON-3, the highest peacetime level it can reach. The US reached this after 9/11, during the Yom Kippur War in 1973, and partially during the Suez Crisis in 1965 (SAC) and during the Cuban Missile Crisis in 1962 (military commands). The highest known level ever reached is DEFCON-2, by SAC, during the Cuban Missile Crisis. This is the level where a local commander could plausibly launch nuclear weapons without authorization. It does not happen very often, and the US has never reached it since 1962 again.

The closest we have come to an accident like this are two incidents, both from the Cuban Missile Crisis. During the crisis, an American U-2 plane on a routine mission to collect air samples over the North Pole drifted into Soviet airspace by mistake, the Soviets detected it, and scrambled fighters to intercept it. The US sent two F-102s fighters to escort it back but because of DEFCON-2, they were carrying nuclear -tipped missiles instead of conventional interceptors. In the event, the U-2 made it out before the Soviets got to it, but one escalation possibility could have been that US fighters engage the Soviet ones, with nukes, over the USSR. When President Kennedy was told of the incident, he remarked that “there’s always some son-of-a-bitch who doesn’t get the word.”

Also during the crisis, the Soviets had their own near-launch, as revealed in 1997 after 35 years of suppressing the story, from their submarine B-59. On October 27, the sub had surfaced to charge its batteries, and was harassed by US ships and planes, which illuminated it with lights, dropped depth charges, and firing warning shots. The commander Savitsky, apparently thinking that they were under attack because a war had broken out, ordered an immediate dive and called for the launch of the nuclear torpedo but was unable to get down the cone to tell the officer in charge of executing the order. His Chief of Staff Arkhipov, who had remained on top realized that the Americans were signaling, not attacking, and called back Savitsky who then signaled to the Americans to cease provocative actions. They did, and the situation never escalated beyond that. The possibility here could have been that a Soviet submarine fires a nuclear weapon at American ships trying to enforce the naval blockade. It is not known whether Khrushchev expressed sentiment similar to Kennedy’s when informed of the incident (if he was ever informed).

One very real possibility during the Cuban Missile Crisis was that an American ground invasion — which had been planned and in motion when the Soviets capitulated — could have caused local Soviet commanders finding themselves under fire to defend the island with tactical nuclear weapons, which they already had (the US did not know that) and which did not require authorization from Moscow to use. It’s possible that one reason Khrushchev backed down when he did is the realization that the Americans might invade the island without realizing the risks they were running.

What are we to make of these cases? The first thing to remember is that they all occurred at a time of intense crisis, arguably the worst confrontation between the superpowers ever experienced. In fact, both sides drew the same conclusion: nuclear brinkmanship is an exceedingly dangerous policy that should not be used. It was never used again (the other significant confrontation was at DEFCON-3 in 1973). The cases occurred when the command-and-control systems were in their infancy, and the risks were correspondingly higher. These days, everything is being carefully monitored, including the storage of nuclear warheads. When the US government says it has detected no signs that Russia is planning to use nuclear weapons, it’s not only because of spies in the Kremlin — of whom we seem to have some — but also because we use satellites to monitor what goes on in Russia, and can find out when preparations for use are underway. This is why I was confident in dismissing the crazy story that Russians are moving nuclear warheads to airports after the drone attacks on the Kremlin: if that had happened, NATO would have lit up like a Christmas tree in an instant. Whatever the antagonism between Moscow and Washington these days is, it is definitely not near the stress level of an actual crisis, and so it is hard to imagine a scenario that would approximate the close-calls we have had.

The second thing to remember is that the frightening tales about accidental war that usually involve retelling of these “near-misses” invariably omit one step, which is absolutely crucial, for a war to take place: a government still has to order the use of nuclear weapons after an initial detonation. That is, the stories induce fear because they go something like this: imagine if a nuke had been used, then… NUCLEAR WAR! But a single detonation does not make a nuclear war, which requires a decision to authorize nukes on a large scale by at least one side. And this is where the problem — or rather, the answer to the problem — lies: why would a government order the launch of nuclear weapons on news of a single explosion when everyone is fully aware of the possibility of accidents or unauthorized launches? They would not. Whatever retaliatory measures are taken, nobody would be unleashing nuclear Armageddon without making triply sure they are under a real attack.

Some might say this is too optimistic, that people in charge might not be rational, and so on. I don’t actually subscribe to the extreme versions of this “explanation” but even if one did, the issue here would no longer be accidental nuclear war but quite a deliberate one.

The upshot of all of this is that “accidental nuclear war” is not a thing. The likelihood of it happening and “things spiraling out of control” is so remote that no reasonable policy-maker should be considering this when formulating policies for the present war. (This does not mean that we should not be testing, and refining, our safety protocols, of course.)

What one does need to consider is the purposeful use of nuclear weapons. Here, I am on record last year for saying that it’s possible to imagine a scenario under which Putin would order the use of a tactical device for political purposes. There was a period of about six months, at least, last year when the Kremlin could have thought that a demonstrative use of a nuclear weapon (not even necessarily in Ukraine) could scare off the West from supporting Ukraine. Even if this was ever seriously considered, Putin must have shelved the idea when he resorted to mobilization in September — the choice to use conventional means of pursuing his goals in Ukraine told me that the Russians must have decided that the costs of using a nuclear weapon far outweigh the gains (again, if they even considered it). I remain convinced that Russia would not use nuclear weapons unless it is in self-defense in an existential crisis, which is not something that it is going to encounter in the current war. I also am not persuaded that Putin would order it because he equates his personal survival with that of Russia. However, as I am not able to peer into his mind, this remains a possibility, and it must be considered. Whatever one’s estimate of this risk is — and, again, I think it is vanishingly small — it is not one of “accidental nuclear war” and the mechanisms involved then are quite different, and a topic perhaps for another post.

One thought on “Accidental Nuclear War: Don’t Lose Sleep over It

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s