[Editors’ note: This essay is one in a series—the Good Governance Papers—organized by Just Security. In these essays, leading experts explore actionable legislative and administrative proposals to promote non-partisan principles of good government, public integrity, and the rule of law. For more information, you can read the Introduction by the series’ editors.]

With the possible exception of President Richard Nixon’s reported boast that he could “pick up the telephone and within 25 minutes millions of people will be dead,” no U.S. president has spoken more flippantly about the use of nuclear weapons than President Donald Trump. And certainly no president has spoken so flippantly so often.

Memorably, Trump in the midst of a crisis with North Korea threatened to rain “fire and fury” upon the paranoid and nuclear-armed country, publicly taunted the North’s young and insecure leader about having “a much bigger & more powerful” nuclear “Button,” reportedly inquired with aides about nuking hurricanes, and challenged Russia to a new arms race. These remarks must be read together with the president’s well-reported impulsive nature, unfamiliarity with national security issues, disinterest in learning about the world, disdain for an orderly decision process, and angry denigration of senior Pentagon leaders. Concern about the President’s cavalier attitude about the bomb has, accordingly, been deep and wide. According to Bob Woodward, during the Korea crisis Trump’s Secretary of Defense James Mattis was sleeping in his clothes. This was after the Secretary reportedly took steps to ensure that either he or another retired general were never far from the president, and that no nuclear order could be transmitted without his knowledge.

For the 75 year history of America’s nuclear arsenal, two cardinal norms have held: first, great care and self-restraint by the Commander-in-Chief in any discussion of nuclear weapons, and second, maintenance of a nuclear command, control, and communications (NC3) system designed as a matter of law, technology, and decision process around the president’s judgment. Trump abandoned the first norm. It is therefore time – indeed long past time – to revisit the second.

Whoever wins the 2020 election, Congress can and should write a decision process statute for nuclear weapons. As I explain in a forthcoming article, this statute would preserve rapid presidential decision in the face of initiated or truly imminent adversary nuclear (or WMD) attack, while also building norms that would reduce the risk of an unwarranted or impulsive launch where the nation does not truly face an imminent threat. The latter can be accomplished through well-established congressional powers of legislating use-of-force criteria, assigning roles, defining terms, creating reporting requirements, and establishing other accountability mechanisms that foster “good process” expectations. Without intruding in any formal way on the president’s constitutional powers, the statute would exercise Congress’ constitutional powers to reduce the risk that a rogue or precipitous president could order an atomic atrocity.

This essay describes the current NC3 system and three enduring nuclear nightmares, explains the need to build stronger norms about review of nuclear decisions, makes the case for a process-creating statute, argues that the constitutional case for a legislated structure has only grown stronger, and outlines the statute I recommend.

Nuclear Command and Control, and Three Enduring Nuclear Nightmares

The NC3 system inherited from the Cold War has been technologically modernized but – for better and for worse – remains functionally unchanged. There is no second vote in the NC3 system, only a chain of command created by the Constitution and by statute that features a secretary of defense appointed by the president and military officers who under the Uniform Code of Military Justice may disregard only unambiguously, patently illegal orders. As I write in my article:

A primary virtue of the system’s legal, decision process, and technology architecture is also a potential liability: concentration of discretion to use the world’s most destructive weapons in one person. What could enable timely presidential decision in the classic nuclear nightmare of an [initiated or temporally imminent] adversary nuclear attack may permit other nightmares: an unwarranted launch order [by a President who has gone rogue], or a precipitous order where the necessity and legality of the strike are questionable and the President has bypassed advisors and ignored pertinent fact and law.

We still need nuclear deterrence and the highly responsive NC3 system that in part undergirds it because the risk that the United States could find itself quite suddenly in a nuclear crisis is low but endures. If anything, the risk of a nuclear crisis with another nuclear-armed state has gone up in recent years, thanks to rising tensions with Russia, China, and North Korea. To deter the classic nuclear nightmare of an attempted adversary decapitation attack, the vast majority of nuclear deterrence experts will insist that the president must retain the ability to make a decision on an unforgiving timeline – the mere minutes it takes a missile to travel from a submarine or foreign soil to the United States.

Trump’s conduct regarding nuclear weapons has demonstrated that the system’s dependence on one person’s soundness of mind involves risk of nuclear nightmares. One is a Rogue President who orders an atomic atrocity while suffering from the kinds of mental health or substance abuse problems that often afflict human beings, including many national leaders throughout history. President Nixon’s intoxicated orders to launch nuclear weapons, for example, reportedly only died because his staff and military leaders decided to wait on transmitting the command until the president sobered up, and the president then forgot about it as he slept off his hangover. (This incident may be apocryphal, but it has some consistency with other accounts of Nixon’s conduct while drunk, and is certainly plausible). The other nuclear nightmare is the Precipitous President, who faces a real but ambiguous gathering threat and reaches too quickly for “the button” when non-nuclear options are clearly available. In both unlikely but entirely possible scenarios, the Constitution’s presidential-removal processes – elections, impeachment, and cabinet removal under the 25th Amendment – are obviously too slow to stop abuse of an NC3 system configured to launch thousands of nuclear weapons within a handful of minutes upon one individual’s order. The only options left to check a rogue or precipitous president are persuasion, resignation, or refusal-until-firing by presidential subordinates who would likely have little or no time to analyze a launch order, must presume orders to be lawful, and are removable by the president at any time.

Use-of-Force Decision-Making at the National Leadership Level

An added problem for civilian and military subordinates of a Rogue or Precipitous President is the downside of a big upside: the modern NC3 system has never actually been used. Unlike Russia’s president, the American president has never reached for the nuclear “football” containing the launch codes and convened a real “threat conference” involving top civilian and military officials, including the Secretary of Defense, Chairman of the Joint Chiefs of Staff, and Commander of the U.S. Strategic Command. There are, consequently, no tested-and-true decision process norms, only the military’s schedule for briefing, advising, and carrying out launch orders.

The public record regarding the NC3 process also does not reflect any requirement for involvement of the Attorney General or any other lawyer at the moment of presidential decision. The law and lawyers could be expected to matter only to the extent that senior officials have been trained in relevant international and U.S. law; that top officials would choose to involve a lawyer in a nuclear threat conference; or that the law of armed conflict has been “baked into” the standing nuclear options in the “black book” inside the nuclear football through planning-stage review by military lawyers. Of course, the president as Chief Executive and Commander in Chief could at will exclude anyone she wants from a nuclear threat conference (be they a Secretary, military commander, or lawyer). The president could also call for execution of a standing nuclear strike option under threat circumstances very different from those analyzed by military lawyers when the option was designed. Dramatically changed circumstances (for example, the nature of the threat, or the risk to non-combatants) absolutely should prompt new legal review of old options. But again, the president can bypass pretty much any person, advice, or information she regards as unhelpful – and a Rogue or Precipitous President is by definition likely going to be hostile to delay or being schooled on the latest intelligence or key principles of law. Humans are social animals with social memories informed by norms and tradition, and often do not find their courage when something improper is happening for the first time.

The good news is that careful review of use of force decisions at the national leadership level and central involvement of lawyers are both well-established regarding conventional operations, creating sound decision-making pathways that can be extended to nuclear decisions. For the past two decades of continual combat operations, decisions about conventional strikes by the president and other top officials, with the benefit of careful analysis of alternatives and their legality through the inter-agency process, have become commonplace.

If “good process” procedures and norms worked in many other decisions at the presidential level, there should be an expectation that they would be followed as well if nuclear use is contemplated under similar circumstances: days or weeks of warning, complicated intelligence and legal questions, a variety of options and implications, and high stakes. “Good process” tends to enhance the quality of decisions, and its norms tend to empower actors who are faithful to the facts, our constitutional values, and the law to stand up to impulsive actors.

The Strengthened Constitutional Case for a Statute

The winner of the 2020 election could implement the decision process structure I recommend via executive order. Such an order may help build good process norms. Any president, however, could waive that order at any time, orally and without public notice. Accordingly, a statute would be far more appropriate.

Just because Congress has not written one to date does not mean it cannot do so, and indeed there is a strong originalist and doctrinal case for the constitutionality of statutory use-of-force decisions. The Framers’ vision was one of shared power by Congress and the President. Article I of the Constitution gives Congress the power to create and structure the armed forces, write “Rules” for their “Government and Regulation,” declare war, write such laws as are necessary and proper to bring into effect the powers of the government, and to control all spending – including a default of elimination of the army unless every elected Congress funds it. As Commander in Chief, the President has direction of the forces Congress provides and governs, and can repel imminent attacks.

The second half of the republic’s history has seen increasingly frequent articulation of an executive power view that Congress can place virtually no limits on the President’s uses of the forces Congress provides, but this viewpoint is not well supported by the originalist evidence or our constitutional history. As David Barron and Marty Lederman wrote in their exhaustive study of collisions of congressional and Commander in Chief powers since the founding, there is “surprisingly little” support for the proposition that the President is free of these kinds of legislative restraints regarding military campaigns. In the founding era, the Supreme Court unanimously struck down President Adams’ order to the Navy during the naval war with France as a violation of statute, and in the two centuries since the Court has repeatedly made clear that the Commander in Chief is subject to the statute.

As I argue in my article updating the constitutional conversation about nuclear weapons since its pause at the Cold War’s end, the case for the constitutionality of a statute regarding nuclear decision-making has grown stronger in recent decades. Firmer footing for nuclear rule-writing has been provided doctrinally by the Supreme Court, which has reaffirmed repeatedly the Youngstown (1952) precedent’s vision of shared power and invalidated a presidential wartime order about foreign fighters as contrary to statute. The facts have changed too: superpower nuclear Armageddon has been replaced as the most likely nuclear use scenario by a less time-urgent counter-terrorism-like limited nuclear strike. That reduces practical need to rely exclusively on presidential authority to repel sudden attacks. What’s also new is compelling recent research that has added to the strong weight of Founding Era and other constitutional history in favor of Congress’s legislative powers. Finally, stronger ground for a nuclear statute is also provided by decades of thick practice precedent (constitutional “gloss”) around national security legislative frameworks that provide statutory guardrails to prevent abuse of presidential power, ensure accountability to the elected representatives of the people, and promote good process (specifically, the covert action statute and Foreign Intelligence Surveillance Act (FISA)).

The Nuclear Forces Control Act

These statutory frameworks along with past force authorizations and other laws provide proven approaches to regulating the national security apparatus that Congress can reconfigure to build legal expectations and process norms intended to bar inappropriate action, facilitate quality decisions, and steel the resolve of subordinates who could face a Rogue or Precipitous President.

To be clear, the statute I recommend does not authorize use of nuclear weapons. It is a criteria-providing and process-structuring law that runs over the top of some other legal authority for use of force (war declaration, statutory authorization of force, or presidential constitutional authority to repel an attack). In the same way, the covert action and FISA statutes continue to define and structure decisions about “secret wars” and surveillance even when some other legal authority authorizes military operations of which these activities could be part.

Specifically, the Nuclear Forces Control Act I recommend – a full draft of which is appended to my article – would exercise clear congressional powers in the following ways. The statute:

  • Conditions appropriations. It bars funding for nuclear use inconsistent with the statute.
  • Creates penalties. The statute borrows from the Posse Comitatus Act and FISA criminal penalties for participation in the statute’s violation. If Congress were concerned about pardon-dangling, it could include a FISA-like civil penalty, too.
  • Conditions use of force. As constitutional “Rules” for nuclear forces, the statute codifies the nuclear use doctrine of the Obama and Trump Administrations: nuclear weapons can be employed only in “extreme circumstances” to defend “vital interests.” Today, an unhinged President could waive this policy at will. Also, nuclear weapons may be used only where some other legal authority exists for their use, use is necessary and legal under the law of armed conflict, and then only in two circumstances: (i) where an adversary nuclear attack is imminent, or (ii) where an adversary attack is not imminent and options and implications have been fully evaluated and nuclear use is the only way to deal with a gathering threat.
  • Defines terms. The statute defines “imminent” temporally: adversary nuclear attack is “reasonably possible within the next 72 hours, based on assessment of adversary capabilities and intentions,” with mere adversary possession of “the bomb” not being enough. This provision combats the elastic understanding of “imminent” employed by the Executive Branch in recent decades. It gives subordinates new ground to resist a President who reaches too quickly for “the button.”
  • Structures the decision process and assigns roles. The statute gives the National Security Council Congress created in 1947 (and which includes the Pentagon leadership) the lead role advising the President on nuclear use in non-imminent adversary threat circumstances, with legal review, and under the coordination of the Vice President. The good process expectations fostered in the mind of the VP would be especially salient because the VP would become President if the President were removed via impeachment or under the 25th Amendment.
  • Requires reports to Congress. Of several reporting requirements, the most significant is a requirement that the Director of National Intelligence, Secretary of Defense, Chairman of the Joint Chiefs of Staff, and Commander of the U.S. Strategic Command – the four subordinates who would know the threat situation the best, and three of whom could transmit a launch order – notify congressional leaders anytime that in their own, independent assessment adversary nuclear attack is imminent. If the President is reaching for the button but an official with the best intelligence cannot in good conscience so report, it creates a process point for that official (and other policymakers and lawmakers) to realize that the President is on the wrong path and they need to push back – dissuade the President, resist bureaucratically in some way, or quit and sound the alarm.
  • Provides interpretive guidance. The statute concludes with provisions meant to complicate legal tactics often used by the Executive Branch to avoid statutory limitations. The statute has a FISA-like exclusivity provision, requires its explicit reference for repeal or amendment, and combats its creative reinterpretation in secret (the secret law problem).

This statutory approach leaves untouched what most scholars would regard as the core of presidential powers that are beyond legislative regulation and would be especially relevant in the face of a sudden foreign nuclear threat. The statute does not direct the military in the field, give Congress a “second vote,” displace the President at the top of the military chain of command, require the President to get the assent of any subordinate for anything the President can legally order, nor give any other Executive Branch official authority to order use of nuclear weapons. The statute also leaves untouched presidential power and the nuclear command system’s responsiveness in the face of imminent or initiated adversary nuclear attack.


Such a statute would create reasonable, minimally intrusive limits and process expectations in service of the vitality of the old norms about “the bomb” – great care and restraint on the part of the president, and availability of a highly responsive nuclear command and control system when it is truly needed. In the wake of presidential bragging about big nuclear buttons and threats of raining “fire and fury” on anxious nuclear-armed dictators, it is no longer reasonable, if it ever were, to think that we will always be blessed with a president with good judgment. Congress should summon its institutional courage and invoke its clear constitutional powers to make nuclear weapons, like covert action and surveillance, statutorily-based special national security capabilities. A White House committed to the future security of the country should itself support such legislation to create checks on a future unhinged Commander-in-Chief. Of course, it may take an impulsive nuclear launch, or nearly one, for Congress to act. But wouldn’t it be better to do reasonable things to prevent a U.S.-perpetrated atomic atrocity than to respond to one?