Now they’re his. When Trump takes office in January, he will have sole authority over more than 7,000 warheads. There is no failsafe. The whole point of U.S. nuclear weapons control is to make sure that the president — and only the president — can use them if and whenever he decides to do so. The one sure way to keep President Trump from launching a nuclear attack, under the system we’ve had in place since the early Cold War, would have been to elect someone else.
When the legal framework for nuclear weapons was developed, the fear wasn’t about irrational presidents but trigger-happy generals. The Atomic Energy Act of 1946, which was passed with President Harry Truman’s signature after nine months of acrimonious congressional hearings, firmly put the power of the atomic bomb in the hands of the president and the civilian components of the executive branch. It was a momentous and controversial law, crafted in the months following the atomic bombings of Hiroshima and Nagasaki, with an eye toward future standoffs with the Soviet Union.
The members of Congress who wrote the law, largely with the backing of the scientists who worked on the Manhattan Project, framed it explicitly as a question of who controls the power to use nuclear weapons: Is dropping an atomic bomb a military act or a political one? If it is inherently political, above and beyond a regular military tactic, then that power could not be entrusted to the military. Ultimately, the president was supposed to be the check against the Pentagon pushing to use nukes more often.
The scientists’ fears were based in their experiences in World War II. Their work under the Army Corps of Engineers and the Army Air Forces left them with a sour taste: Generals, they concluded, cared little about ethics, democracy or international politics. Even during the war, some civilians involved with atomic-bombing work feared that the military had become too eager to leave German and Japanese cities in cinders. The secretary of war, Henry Stimson, learned about the ruinous firebombing of Tokyo from the press. He warned Truman that letting the military run the show might cause the United States to “get the reputation of outdoing Hitler in atrocities.”
This division between military and civilian control over nuclear weapons has been weaker or stronger at various points. In the late 1940s, U.S. nuclear weapons could have their nuclear components — the plutonium or uranium “pits” needed to start their reactions — removed and inserted as needed. The nuclear parts of the atomic bombs were in the custody of the civilian Atomic Energy Commission (the precursor to the present-day Energy Department), while the military controlled the nonnuclear parts. The president had the power to transfer these pits to the military and order their use.
During the Eisenhower administration, more compact and complex weapons were developed whose nuclear and nonnuclear parts could not be separated. Fearing a Soviet sneak attack, President Dwight D. Eisenhower put the military in charge of most of the U.S. nuclear stockpile to streamline a possible response. Eisenhower also “pre-delegated” authority to the military to use tactical nuclear weapons (aimed at tanks, not cities) without getting specific presidential approval in certain situations, such as if Soviet tank columns rolled into Germany’s Fulda Gap.
Fears of low-level commanders setting off nuclear conflagrations during the tensions of the early 1960s persuaded President John F. Kennedy to dial some of this back. Miscommunications during the Cuban missile crisis almost led to the use of nuclear weapons by both U.S. and Soviet troops, and U.S. weapons stationed abroad, such as the Jupiter missiles in Turkey, could be used by any army that seized control of them. There were also lingering concerns about “Strangelove”-esque rogue generals. The head of the Strategic Air Command, Gen. Thomas Power, was an enthusiastic proponent of preemptive nuclear war.
Similar concerns within the upper reaches of the Kennedy administration led to a push for technologies to “lock” the nuclear weapons and prevent their use without some kind of codes or authorization. Some early versions were as primitive as combination locks, but later versions were complex electro-mechanical systems that could physically disable a weapon if it were tampered with or if the wrong code was entered too many times.
Eventually, the brass adopted the idea that, when it came to nuclear matters, they were at the beck and call of the president. It was not generals’ responsibility to make the order; it was their responsibility to carry it out.
That the president would be the only person competent to use nuclear weapons was never challenged. Even asking the question would throw the entire system into disarray, as Maj. Harold Hering learned in 1973. Hering was a 21-year Air Force veteran who was decorated for his flying in Vietnam before being sent for training as a nuclear missile squadron commander. He had been taught that officers had an obligation to disobey illegal orders. So when he was told how to launch a nuclear attack, he asked what seemed like a simple question: How could he be sure that an order to launch his missiles was lawful? How could he be sure, for example, that the president wasn’t insane? Instead of an answer, he got the boot: an aborted promotion and an administrative discharge for “failure to demonstrate acceptable qualities of leadership” and for indicating “a defective mental attitude towards his duties.”
The Air Force’s problem, in short, is that once a serviceman starts down the rabbit hole of doubt, he becomes an unreliable second-guesser — and suddenly he is one of the few people who can decide whether nuclear weapons are used.
The procedure for ordering a nuclear attack involves more than one person: The president cannot literally press a button on his desk and start World War III. There is no “nuclear button” at all. Instead, the U.S. nuclear command-and-control system is bureaucratically and technically complex, stretching out to encompass land-based missile silos, submarine-based ballistic and cruise missiles, and weapons capable of being dropped from bombers. The chain of command as widely understood requires that the president order the secretary of defense to carry out a launch; the secretary serves as the conduit for implementation by the military. There are succession policies in place so that the procedure can be continued in the event of the death or incapacitation of either the president or the secretary of defense — or their designated successors.
Most details of how a nuclear war would be started are classified, because an enemy who knew enough about the system could come up with ways to complicate or defeat it. What is known is that an aide is always following the president, carrying at least one large satchel (often two) known as the “nuclear football,” reportedly containing information about nuclear attack possibilities and how the president could verify his identity, authenticate orders and communicate with the military about implementing them.
Could the secretary of defense refuse to carry out a presidential order for a nuclear attack? The legal and constitutional aspects are not clear. The official doctrine that has been released says nothing about this question, and the cryptic public responses to official inquiries, even from Congress, indicate that it is not something that can be openly talked about. “Only the president can authorize the use of nuclear weapons” is essentially the only reply officials ever give to any questions about nuclear controls. Could the president simply fire the defense secretary and move on to the deputy secretary, the secretary of the Army and so on through the chain of command? Maybe. Such an action would at least slow things down, even if the refusal to carry out the order was illegal. But the secretary may not even be formally required to participate — U.S. Air Force doctrine does not indicate he is a necessary part of the chain of command, and holds that the president can communicate directly with the military, in the form of the Joint Chiefs of Staff, to order a nuclear strike.
Commanders further down the pipeline are trained to act quickly on any orders that do come in. The launch officers are trained to launch weapons, not to debate the legality or advisability of the action. Hence the problem with Hering’s question in 1973: While nuclear launch officers are not meant to be strictly mechanical (and indeed, the United States has always resisted fully automating the process), if they stopped to question whether their authenticated orders were legitimate, they would put the credibility of U.S. nuclear deterrence at risk.
Congress held hearings on these issues in the mid-1970s, but nothing came of them. The debate faded away except among a small circle of nuclear wonks. In the early 1980s, Jeremy Stone, then the president of the Federation of American Scientists, proposed that Congress ought to pass a law restricting presidential use of nuclear weapons. The idea was fairly simple: So long as no nuclear weapons had been used by another power in a conflict, the president should not be able to order a first strike with nuclear weapons without getting approval from a fairly large committee of high-ranking members of Congress. It would not eliminate the possibility of an American first strike but would spread the responsibility more democratically.
The idea was pooh-poohed by legal scholars, who noted that Congress has often been far more belligerent than presidents and that the logistics could be complicated.
The people who set up the current command-and-control system did believe there was a check in place: elections. Don’t want an insane president to have nuclear weapons? Don’t put one in office. But this isn’t necessarily much of a check — even rational presidents have bad days; even high-functioning people succumb to mental illness or substance abuse.
It might be worth resurrecting this debate , if we take seriously the idea that presidents — any of them, much less Trump — should not have the legal authority to conduct arbitrary and unilateral nuclear war. Perhaps now, decades after the end of the Cold War, we are past the moment when we need to entrust that power in a single person. One can imagine a law that would allow the president to use nuclear weapons in the face of imminent danger, the sort of situation in which a matter of minutes or even seconds could make a difference, but would enact formal requirements for outside consensus when more options were on the table. It would not require a full renunciation of the possibility of a first-strike nuclear attack (something the United States has never been willing to make) but might add some reassurances that such decisions would not be made unilaterally.
Congress ceded a considerable amount of power to the presidency in 1946. Seventy years later, maybe it is time lawmakers took some of it back.
This story was updated after its initial publication to reflect later research by the writer that found that the secretary of defense may not be legally required to be part of the chain of command for launching nuclear missiles.